notebooks/temporal_action_segmentation_tf.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# Temporal Action Segmentation in Videos Using Recurrent Neural Networks\n",
    "\n",
    "Welcome! In this lab, you'll learn how to train a recurrent neural network on the [Breakfast dataset](http://serre-lab.clps.brown.edu/resource/breakfast-actions-dataset/) for the task of temporal action segmentation and recognition. Our goal is to predict one action label for each frame of a long untrimmed video showing a person preparing breakfast. We will go through all the steps, including loading the data, building and training a model, making per frame label predictions and visualizing results. We will use the [Tensorflow library](https://github.com/tensorflow/tensorflow) to build and train our model."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Runtime\n",
    "Before we start, make sure you are using a GPU Runtime on Google Colab. (Runtime -> Change Runtime -> GPU)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Breakfast Dataset Overview\n",
    "\n",
    "[Breakfast dataset](http://serre-lab.clps.brown.edu/resource/breakfast-actions-dataset/) is a video dataset containing 1712 long untrimmed videos, each one containing a sequence of multiple actions. Videos contain actions related to breakfast preparation, performed by 52 different individuals in 18 different kitchens. It features videos from multiple cameras, which were uncalibrated and their position changes based on the location. There are ∼77 hours of video (> 4 million frames). The resolution of videos is 320×240 pixels and the frame rate is 15 fps.\n",
    "\n",
    "For this lab, we will use precomputed features describing each frame of the video and our goal is to predict a label for the action performed by the person in each frame of a video.\n",
    "![Example output](http://serre-lab.clps.brown.edu/wp-content/uploads/2012/04/juice_frame550.png)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Uncomment for google collaboratory\n",
    "# !git clone https://github.com/emavroudi/jsalt18-actrec-lab\n",
    "# !mkdir /content/jsalt18-actrec-lab/data/breakfast/feat  \n",
    "# !wget -P /content/jsalt18-actrec-lab/data/breakfast/feat http://www.vision.jhu.edu/workshop/jsalt2018/dt_l2pn_c64_pc64.tar.gz\n",
    "# !tar -xzf jsalt18-actrec-lab/data/breakfast/feat/dt_l2pn_c64_pc64.tar.gz -C jsalt18-actrec-lab/data/breakfast/feat/\n",
    "# !ls -1 jsalt18-actrec-lab/data/breakfast/feat/dt_l2pn_c64_pc64/split00 | wc -l\n",
    "# import os\n",
    "# os.chdir('./jsalt18-actrec-lab/notebooks')\n",
    "# !pwd\n",
    "# import sys\n",
    "# sys.path.append('/content/jsalt18-actrec-lab')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Import libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/cis/home/efi/work/ar_libs/anaconda3/envs/action_detection_tf1.6/lib/python3.6/site-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
      "  from ._conv import register_converters as _register_converters\n"
     ]
    }
   ],
   "source": [
    "from __future__ import absolute_import\n",
    "from __future__ import division\n",
    "from __future__ import print_function\n",
    "\n",
    "\n",
    "import logging\n",
    "from numpy import array, vstack, sum, prod, argmax, nonzero, mean\n",
    "import os\n",
    "import time\n",
    "from collections import namedtuple, defaultdict\n",
    "import pdb\n",
    "from IPython.display import Image\n",
    "\n",
    "\n",
    "import tensorflow as tf\n",
    "from tensorflow.python.ops import clip_ops, array_ops, nn_ops, math_ops\n",
    "from tensorflow.python.framework import ops\n",
    "from tensorflow.contrib.rnn import LSTMCell, GRUCell, MultiRNNCell\n",
    "\n",
    "from datasets.breakfast import BreakfastDataset\n",
    "from datasets.batch_generator import FrameSequenceBatchGenerator\n",
    "from utils.preprocessing import pad_sequences, DataPreprocessor\n",
    "from utils.metrics import per_frame_accuracy\n",
    "from utils.tf_utils import debug_nans, to_categorical, clip_gradients, \\\n",
    "    stack_bidirectional_dynamic_rnn\n",
    "from utils.plot_utils import plot_optimization_log_frame, viz_sequence_predictions, plot_label_seq\n",
    "from utils.my_io_utils import save_to_pickle\n",
    "from utils.misc import frame_labels_to_segments"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 1) Load data\n",
    "\n",
    "We will now load precomputed features from training and testing videos of the Breakfast dataset. Each video frame is described by a $64$ dimensional feature vector. These are hand-crafted features extracted by the authors of the dataset and are [publicly available](https://uni-bonn.sciebo.de/index.php/s/lqj3GNVQWXaX1WC). For computational efficiency and easier training, we downsample videos in time, i.e. we use only 1 frame per second."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "training_annotations_json_file = '../data/breakfast/annotations_train_split0.json'\n",
    "testing_annotations_json_file = '../data/breakfast/annotations_test_split0.json'\n",
    "downsampling_factor = 15 # Downsample \n",
    "\n",
    "# BreakfastDataset class implements methods: __len__(): returning the number of training/testing samples/videos\n",
    "# and __getitem__: returning a dictionary with keys: 'video_name', 'feat', 'labels', 'frame_indices'\n",
    "training_set = BreakfastDataset(dataset_path='../data/breakfast', downsampling_factor=downsampling_factor,\n",
    "                 annotations_json_file=training_annotations_json_file)\n",
    "\n",
    "testing_set = BreakfastDataset(dataset_path='../data/breakfast', downsampling_factor=downsampling_factor,\n",
    "                 annotations_json_file=testing_annotations_json_file)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Let's look into one training sample"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Nb training videos: 1423\n"
     ]
    }
   ],
   "source": [
    "nb_training_samples = len(training_set)\n",
    "print('Nb training videos: {:d}'.format(nb_training_samples))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "training_sample_ind = 0\n",
    "training_sample = training_set[training_sample_ind]\n",
    "segs, seg_labels = frame_labels_to_segments(training_sample['labels'])\n",
    "action_names = training_set.get_action_names()\n",
    "action_names_sequence = [action_names[seg_label] for seg_label in seg_labels]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training sample 0 video name: P03_cam01_P03_cereals\n"
     ]
    }
   ],
   "source": [
    "print('Training sample {} video name: {}'.format(training_sample_ind, training_sample['video_name']))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training sample 0 feature shape (nb_frames, feat_dim): (56, 64)\n"
     ]
    }
   ],
   "source": [
    "print('Training sample {} feature shape (nb_frames, feat_dim): {}'.format(training_sample_ind, str(training_sample['feat'].shape)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training sample 0 per frame labels: [ 0  0 39 39 39 39 39 39 39 39 11 11 11 11 11 11 11 11 11 11 11 11 11 11\n",
      " 11 11 11 11 11 17 17 17 17 17 17 17 17 17 17 31 31 31 31 31 31 31 31  0\n",
      "  0  0  0  0  0  0  0  0]\n"
     ]
    }
   ],
   "source": [
    "print('Training sample {} per frame labels: {}'.format(training_sample_ind, str(training_sample['labels'])))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training sample 0 sequence of actions : ['SIL', 'take_bowl', 'pour_cereals', 'pour_milk', 'stir_cereals', 'SIL']\n"
     ]
    }
   ],
   "source": [
    "print('Training sample {} sequence of actions : {}'.format(training_sample_ind, str(action_names_sequence)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training sample 0 frame indices: [  0  15  30  45  60  75  90 105 120 135 150 165 180 195 210 225 240 255\n",
      " 270 285 300 315 330 345 360 375 390 405 420 435 450 465 480 495 510 525\n",
      " 540 555 570 585 600 615 630 645 660 675 690 705 720 735 750 765 780 795\n",
      " 810 825]\n"
     ]
    }
   ],
   "source": [
    "print('Training sample {} frame indices: {}'.format(training_sample_ind, str(training_sample['frame_indices'])))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Action classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "SIL\n",
      "add_saltnpepper\n",
      "add_teabag\n",
      "butter_pan\n",
      "crack_egg\n",
      "cut_bun\n",
      "cut_fruit\n",
      "cut_orange\n",
      "fry_egg\n",
      "fry_pancake\n",
      "peel_fruit\n",
      "pour_cereals\n",
      "pour_coffee\n",
      "pour_dough2pan\n",
      "pour_egg2pan\n",
      "pour_flour\n",
      "pour_juice\n",
      "pour_milk\n",
      "pour_oil\n",
      "pour_sugar\n",
      "pour_water\n",
      "put_bunTogether\n",
      "put_egg2plate\n",
      "put_fruit2bowl\n",
      "put_pancake2plate\n",
      "put_toppingOnTop\n",
      "smear_butter\n",
      "spoon_flour\n",
      "spoon_powder\n",
      "spoon_sugar\n",
      "squeeze_orange\n",
      "stir_cereals\n",
      "stir_coffee\n",
      "stir_dough\n",
      "stir_egg\n",
      "stir_fruit\n",
      "stir_milk\n",
      "stir_tea\n",
      "stirfry_egg\n",
      "take_bowl\n",
      "take_butter\n",
      "take_cup\n",
      "take_eggs\n",
      "take_glass\n",
      "take_knife\n",
      "take_plate\n",
      "take_squeezer\n",
      "take_topping\n"
     ]
    }
   ],
   "source": [
    "print('\\n'.join(action_names))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 2) Preprocess data\n",
    "\n",
    "Preprocessing includes padding feature sequences, so that they all have the same length (corresponding to the maximum length of all sequences), and converting labels from an integer format (e.g., \"2\"), to an [one hot encoding](https://en.wikipedia.org/wiki/One-hot) (e.g., \"0, 0, 1, 0, 0, 0, 0, 0, 0, 0\"). Therefore, after this preprocessing each batch of data will be described by features of shape: `(batch_size, max_nb_frames, 64)` and labels of shape: `(batch_size, max_nb_frames, 10)`. \n",
    "\n",
    "To achieve this, we define a class that implements a preprocess method with arguments features_lst and frame_labels_lst. Later, we will use the variable `sample_frame_weights` to ignore padded frames."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "class RNNDataPreprocessor(DataPreprocessor):\n",
    "    def __init__(self, preprocessor_params):\n",
    "        \"\"\"\n",
    "        :param preprocessor_params: dict with keys:\n",
    "                       'encoder_nb_classes' and 'max_nb_frames',\n",
    "                       encoder_nb_classes: number of classes for encoder\n",
    "                                          (number of action labels)\n",
    "                       max_nb_frames: maximum number of frames\n",
    "\n",
    "        \"\"\"\n",
    "        super().__init__(preprocessor_params)\n",
    "\n",
    "    def preprocess(self, features_lst, frame_labels_lst):\n",
    "        \"\"\"\n",
    "        Pads feature sequences and frame_labels to max_nb_frames\n",
    "        Converts frame_labels to one-hot encoding\n",
    "\n",
    "        :param features_lst: list with nb_samples elements. Each element\n",
    "                              is a numpy array of size (timesteps, feat_dim)\n",
    "        :param frame_labels_lst: a list with nb_samples elements.\n",
    "                    Each element is a (nb_timesteps,) numpy array or list,\n",
    "                    each element of which list is a list of the labels for\n",
    "                    this timestep\n",
    "        :return: feat: numpy array with shape (nb_samples, max_nb_frames,\n",
    "                       feat_dim)\n",
    "                 frame_labels: numpy array with shape\n",
    "                              (nb_samples, max_nb_frames, encoder_nb_classes)\n",
    "                 frame_sample_weights: numpy binary array with shape\n",
    "                   (nb_samples, max_nb_frames, 1),\n",
    "                   having zeroes at padded indices and ones elsewhere\n",
    "                 frame_sequence_lengths: numpy array with shape = (nb_samples, )\n",
    "                   with true length (nb_timesteps) per sample sequence\n",
    "        \"\"\"\n",
    "        encoder_nb_classes = self.params['encoder_nb_classes']\n",
    "        max_nb_frames = self.params['max_nb_frames']\n",
    "\n",
    "        # ----------------------------------------------------------------------\n",
    "        # Process features and frame_labels (Pad and one-hot encoding)\n",
    "        frame_sequence_lengths = [x.shape[0] for x in features_lst]\n",
    "        feat, frame_sample_weights = pad_sequences(\n",
    "            features_lst, max_len=max_nb_frames,\n",
    "            dtype='float32', value=0.)\n",
    "\n",
    "        # Format labels\n",
    "        # Go from y_t = {1...C} to one-hot vector (e.g. y_t = [1, 0, 1, 0])\n",
    "        # list of arrays (timesteps, num_classes))\n",
    "        frame_labels_lst = [to_categorical(y, encoder_nb_classes)\n",
    "                            for y in frame_labels_lst]\n",
    "\n",
    "        # Pad frame labels sequences\n",
    "        frame_labels, _ = pad_sequences(frame_labels_lst, max_len=max_nb_frames,\n",
    "                                        dtype='float32', value=0.)\n",
    "\n",
    "        return feat, frame_labels, frame_sample_weights, \\\n",
    "            frame_sequence_lengths"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "nb_classes = len(action_names)\n",
    "preprocessor_obj = RNNDataPreprocessor(\n",
    "    preprocessor_params={\n",
    "        'encoder_nb_classes': nb_classes,\n",
    "    })"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3) Define data generators, providing batches of data\n",
    "\n",
    "We will now get training and testing data generators, implementing the method __next__(), which is yielding batches of data. Each batch contains features: `(batch_size, max_nb_frames, feat_dim)`,  frame_labels: `(batch_size, max_nb_frames, nb_classes)`, frame_sample_weights: `(batch_size, max_nb_frames,)` and frame_sequence_lengths: `(batch_size,)`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "batch_size = 32\n",
    "\n",
    "training_batch_generator_obj = FrameSequenceBatchGenerator(batch_size, dataset_obj=training_set, preprocessor_obj=preprocessor_obj, nb_classes=nb_classes,\n",
    "                 shuffle=True, seed=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "testing_batch_generator_obj = FrameSequenceBatchGenerator(batch_size, dataset_obj=testing_set, preprocessor_obj=preprocessor_obj, nb_classes=nb_classes,\n",
    "                 shuffle=False, seed=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Maximum number of frames in training videos: 650\n",
      "Maximum number of frames in testing videos: 557\n"
     ]
    }
   ],
   "source": [
    "max_nb_frames_train = training_batch_generator_obj.get_max_nb_frames()\n",
    "print('Maximum number of frames in training videos: {:d}'.format(max_nb_frames_train))\n",
    "max_nb_frames_test = testing_batch_generator_obj.get_max_nb_frames()\n",
    "print('Maximum number of frames in testing videos: {:d}'.format(max_nb_frames_test))\n",
    "max_nb_frames = max(max_nb_frames_train, max_nb_frames_test)\n",
    "training_batch_generator_obj.configure_preprocessing(_max_nb_frames=max_nb_frames)\n",
    "testing_batch_generator_obj.configure_preprocessing(_max_nb_frames=max_nb_frames)\n",
    "\n",
    "feat_dim = training_batch_generator_obj.get_feat_dim()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 4) Build tensorflow model\n",
    "\n",
    "We will now build our Recurrent Neural Network model. TensorFlow uses a dataflow graph to represent our computation in terms of the dependencies between individual operations. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Recurrent Neural Networks\n",
    "Recurrent Neural Networks (RNNs) are a family of neural networks for processing sequential data. The recurrent connection allows the propagation of information through time. They have a \"memory\" that keeps tracks of information observed so far. This memory is captured in the hidden states of the network.\n",
    "\n",
    "The hidden state at time $t$ is calculated based on the previous hidden state and the input at the current step, i.e.\n",
    "$h_t = f(h_{t-1}, x_t;W, b)$, where $h_t$ is called the `hidden state` at time $t$, $x_t$ is the `input` at time $t$ (in our case the 64-dimensional feature vector describing frame $t$) and $f(\\cdot)$ is a nonlinear function with trainable weights and biases $W, b$.\n",
    "The output at time $t$ (in our case scores per action class) is computed based on the hidden state at time $t$. $y_t = W_{hy} h_t + b_{y}$\n",
    "<img src=\"https://cdn-images-1.medium.com/max/1600/1*4KwIUHWL3sTyguTahIxmJw.png\" alt=\"RNN\" height=\"200px\" width=\"500px\"/>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Bidirectional Recurrent Neural Networks\n",
    "Bidirectional Recurrent Neural Networks integrate information from the future as well as the past to make a prediction for each frame in the video. Therefore, they are expected to be better at predicting the temporal boundaries of an action as compared to a unidirectional RNN. The hidden states of the forward and backward RNNs are usually concatenated at each time step.\n",
    "\n",
    "$\\overrightarrow{h_t} = f(\\overrightarrow{h_{t-1}}, x_t;\\overrightarrow{W}, \\overrightarrow{b})$\n",
    "\n",
    "$\\overleftarrow{h_t} = f(\\overleftarrow{h_{t-1}}, x_t;\\overleftarrow{W}, \\overleftarrow{b})$\n",
    "\n",
    "$y_t = W_{hy}[\\overrightarrow{h_t}; \\overleftarrow{h_t}] + b_y$\n",
    "<img src=\"https://cdn-images-1.medium.com/max/1600/1*gpBBcQbdtIceERjJ784j1A.png\" alt=\"BiLSTM\" height=\"200px\" width=\"400px\"/>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Multilayer Recurrent Neural Networks\n",
    "\n",
    "We can build multilayer RNNs by stacking RNN layers on top of each other. The hidden states of RNN below are taken as inputs to the RNN above. We can also add residual connections between RNN layers.\n",
    "For instance, stacking bidirectional RNNs can be summarized in the following equations:\n",
    "$\\overrightarrow{h_t^{(j)}} = f(\\overrightarrow{h_{t-1}^{(j)}}, h_t^{(j-1)};\\overrightarrow{W^{(j)}}, \\overrightarrow{b^{(j)}}), \\quad j=1, \\ldots$\n",
    "\n",
    "$\\overleftarrow{h_t^{(j)}} = f(\\overleftarrow{h_{t-1}^{(j)}}, h_t^{(j-1)};\\overleftarrow{W^{(j)}}, \\overleftarrow{b^{(j)}})\\quad j=1, \\ldots$\n",
    "\n",
    "$h_t^{(0)} = x_t$ and $h_t^{(j)} = [\\overrightarrow{h_t{(j)}}; \\overleftarrow{h_t{(j)}}], \\quad j=1,\\ldots$\n",
    "\n",
    "<img src=\"https://i.pinimg.com/originals/2d/f5/ee/2df5eecbda603de43c3de531e431f31e.png\" alt=\"BiLSTM\" height=\"200px\" width=\"200px\"/>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Gated Recurrent Unit (GRU)\n",
    "\n",
    "Gated Recurrent Unit is a particular instance of the hidden state update function $f$ defined above and it uses a gating activation function.\n",
    "\n",
    "$r_t = \\sigma_r(W_r x_t + U_r h_{t-1} + b_r)$ (Update gate)\n",
    "\n",
    "$z_t = \\sigma_z(W_z x_t + U_z h_{t-1} + b_z)$ (Reset gate)\n",
    "\n",
    "$\\tilde{h_t} = \\tanh(W x_t + r_t \\odot (Uh_{t-1}) + b)$ (New memory)\n",
    "\n",
    "$h_t = (1-z_t) \\odot h_{t-1} + z_t \\odot \\tilde{h_t}$ (Hidden state)\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4a) First, we set the values of our hyperparameters."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "DEBUG = 0\n",
    "\n",
    "# Output returned after running a forward pass through our model.\n",
    "InferenceOutput = namedtuple(\n",
    "    \"InferenceOutput\",\n",
    "    \"frame_logits frame_predictions frame_y_pred\")\n",
    "\n",
    "# Results on validation set\n",
    "ValFrameResults = namedtuple(\n",
    "    \"ValFrameLevelResults\",\n",
    "    \"val_frame_metric val_frame_loss val_frame_predictions val_frame_y_pred \"\n",
    "    \"val_frame_y_true val_frame_sample_weights \"\n",
    "    \"val_frame_logits\"\n",
    ")\n",
    "\n",
    "# Results on training set\n",
    "TrainFrameResults = namedtuple(\n",
    "    \"TrainFrameLevelResults\",\n",
    "    \"train_frame_metric train_frame_loss train_frame_predictions \"\n",
    "    \"train_frame_y_pred train_frame_y_true train_frame_sample_weights \"\n",
    "    \"train_frame_logits\"\n",
    ")\n",
    "\n",
    "# Output of rnn encoder\n",
    "EncoderOutput = namedtuple(\n",
    "    \"EncoderOutput\",\n",
    "    \"outputs final_state\")\n",
    "\n",
    "# User defined params\n",
    "params = {\n",
    "    \"feat_dim\": feat_dim,\n",
    "    \"encoder_nb_classes\": nb_classes,\n",
    "    \"max_nb_frames\": max_nb_frames,\n",
    "    # Model parameters\n",
    "    \"encoder_nb_hidden_units\": 256,\n",
    "    \"encoder_cell_type\": 'gru',  # or lstm\n",
    "    \"encoder_is_bidirectional\": True,\n",
    "    \"encoder_activation\": 'tanh',\n",
    "    \"encoder_initializer\": 'random_uniform',\n",
    "    \"encoder_init_scale\": 0.1,\n",
    "    \"encoder_clip_gradients\": 1,\n",
    "    \"encoder_nb_layers\": 2,\n",
    "    # Training/Optimizer parameters\n",
    "    \"dropout_rate\": 0.3,\n",
    "    \"optimizer_name\": \"Adam\",\n",
    "    \"learning_rate\": 0.001,\n",
    "    \"nb_epochs\": 25,\n",
    "    \"shuffle\": True,\n",
    "    \"momentum\": 0.0,\n",
    "    \"decay_period\": 10.0,\n",
    "    \"decay_rate\": 0.5,\n",
    "    \"nesterov\": False,\n",
    "    \"seed\": 42,\n",
    "    \"log_dir\": '../data/breakfast/results',\n",
    "}\n",
    "\n",
    "if params['encoder_initializer'] == 'None':\n",
    "    params['encoder_initializer'] = None\n",
    "if params['encoder_clip_gradients'] == -1:\n",
    "    params['encoder_clip_gradients'] = None\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4b) Then, we define our network architecture\n",
    "The function `_inference()` will build all the operations required to go from an input sequence of features to an output sequence of class predictions per frame."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "def output_fc_layer(layer_inputs, initializer):\n",
    "    # Output Fully Connected layer\n",
    "    layer_name = \"encoder_time_fully_connected\"\n",
    "    with tf.name_scope(layer_name):\n",
    "        # Apply dense layer to each timestep\n",
    "        # kernel: (nb_hidden_states , nb_classes)\n",
    "        # bias: (nb_classes,)\n",
    "\n",
    "        unstacked = tf.unstack(layer_inputs, axis=1)\n",
    "        dense_res = [tf.layers.dense(\n",
    "            inputs=s, units=params['encoder_nb_classes'],\n",
    "            activation=None,\n",
    "            use_bias=True, kernel_initializer=initializer,\n",
    "            bias_initializer=tf.zeros_initializer(),\n",
    "            kernel_regularizer=None, bias_regularizer=None,\n",
    "            activity_regularizer=None, trainable=True,\n",
    "            name=layer_name, reuse=(i != 0))\n",
    "            for (i, s) in enumerate(unstacked)]\n",
    "\n",
    "        # output: 3D Tensor (batch_size, nb_timesteps,\n",
    "        # nb_classes)\n",
    "        frame_logits = tf.stack(dense_res, axis=1)\n",
    "        frame_logits = debug_nans(frame_logits, \"frame_logits\", debug=DEBUG)\n",
    "        print(\"Logits static shape: \",\n",
    "              frame_logits.shape)\n",
    "        frame_predictions = tf.nn.softmax(frame_logits)\n",
    "        frame_y_pred = tf.cast(tf.argmax(frame_logits, axis=-1), tf.int32)\n",
    "\n",
    "        # Create summaries for tensorboard visualization\n",
    "        fc_vars = tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES,\n",
    "                                    layer_name)\n",
    "        tf.summary.histogram('kernel', fc_vars[0])\n",
    "        tf.summary.histogram('bias', fc_vars[1])\n",
    "        tf.summary.histogram('frame_logits', frame_logits)\n",
    "        tf.summary.histogram('frame_predictions', frame_predictions)\n",
    "\n",
    "        return frame_logits, frame_predictions, frame_y_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_rnn_cell(rnn_cell_params, is_training, layer_name='rnn_cell'):\n",
    "    with tf.variable_scope(layer_name):\n",
    "\n",
    "        # Initialize cell weights/biases\n",
    "        if rnn_cell_params['initializer'] is None:\n",
    "            initializer = None\n",
    "        elif rnn_cell_params['initializer'] == \"random_uniform\":\n",
    "            init_scale = rnn_cell_params['init_scale'] / rnn_cell_params[\n",
    "                'nb_hidden_units']\n",
    "            initializer = tf.random_uniform_initializer(\n",
    "                -init_scale, init_scale)\n",
    "        elif rnn_cell_params['initializer'] == 'random_normal':\n",
    "            initializer = tf.random_normal_initializer(\n",
    "                0, rnn_cell_params['init_scale'])\n",
    "        else:\n",
    "            raise ValueError(\"Invalid initializer %s\",\n",
    "                             rnn_cell_params['initializer'])\n",
    "        \n",
    "        # Get forward and backward cells for bidirectional RNN. Weights are not shared.\n",
    "        if rnn_cell_params['cell_type'] == 'lstm':\n",
    "            # LSTM Cell\n",
    "            # Default activation: tanh\n",
    "            if rnn_cell_params['activation'] == 'tanh':\n",
    "                activation = tf.nn.tanh\n",
    "            else:\n",
    "                activation = None\n",
    "\n",
    "            cell_fw = LSTMCell(num_units=rnn_cell_params['nb_hidden_units'],\n",
    "                               use_peepholes=False,\n",
    "                               cell_clip=None, initializer=initializer,\n",
    "                               num_proj=None, proj_clip=None,\n",
    "                               forget_bias=1.0, state_is_tuple=True,\n",
    "                               activation=activation)\n",
    "            cell_bw = LSTMCell(num_units=rnn_cell_params['nb_hidden_units'],\n",
    "                               use_peepholes=False,\n",
    "                               cell_clip=None, initializer=initializer,\n",
    "                               num_proj=None, proj_clip=None,\n",
    "                               forget_bias=1.0, state_is_tuple=True,\n",
    "                               activation=activation)\n",
    "        elif rnn_cell_params['cell_type'] == 'gru':\n",
    "            # GRU Cell\n",
    "            # If bias_initializer is None, then it starts with bias\n",
    "            # of 1.0\n",
    "            # to not reset and not update.\n",
    "            # Default activation: tanh\n",
    "            if rnn_cell_params['activation'] == 'tanh':\n",
    "                activation = tf.nn.tanh\n",
    "            else:\n",
    "                activation = None\n",
    "            cell_fw = GRUCell(num_units=rnn_cell_params['nb_hidden_units'],\n",
    "                              activation=activation,\n",
    "                              reuse=None,\n",
    "                              kernel_initializer=initializer,\n",
    "                              bias_initializer=None)\n",
    "            cell_bw = GRUCell(num_units=rnn_cell_params['nb_hidden_units'],\n",
    "                              activation=activation,\n",
    "                              reuse=None,\n",
    "                              kernel_initializer=initializer,\n",
    "                              bias_initializer=None)\n",
    "        else:\n",
    "            raise ValueError('Not supported cell type: %s',\n",
    "                             rnn_cell_params['cell_type'])\n",
    "\n",
    "        # Add dropout\n",
    "        output_keep_prob = 1 - rnn_cell_params['dropout_rate']\n",
    "        dropout_cell_fw = tf.contrib.rnn.DropoutWrapper(\n",
    "            cell_fw, input_keep_prob=1.0,\n",
    "            output_keep_prob=tf.cond(\n",
    "                is_training,\n",
    "                lambda: tf.constant(output_keep_prob),\n",
    "                lambda: tf.constant(1.0)),\n",
    "            state_keep_prob=1.0,\n",
    "            variational_recurrent=True, input_size=None,\n",
    "            dtype=tf.float32, seed=None)\n",
    "        dropout_cell_bw = tf.contrib.rnn.DropoutWrapper(\n",
    "            cell_bw, input_keep_prob=1.0,\n",
    "            output_keep_prob=tf.cond(\n",
    "                is_training,\n",
    "                lambda: tf.constant(output_keep_prob),\n",
    "                lambda: tf.constant(1.0)),\n",
    "            state_keep_prob=1.0,\n",
    "            variational_recurrent=True, input_size=None,\n",
    "            dtype=tf.float32, seed=None)\n",
    "\n",
    "        return dropout_cell_fw, dropout_cell_bw"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "def rnn_encoder(input_sequences, is_training, sequence_lengths, encoder_params):\n",
    "    # layer_inputs: [batch_size, max_len, feat_dim]\n",
    "    layer_inputs = input_sequences\n",
    "    layer_inputs = debug_nans(layer_inputs, \"layer_inputs\", debug=DEBUG)\n",
    "    print(\"Layer inputs static shape: \", layer_inputs.shape)\n",
    "    layer_name = \"encoder_rnn\"\n",
    "\n",
    "    # Get forward/backward cells for each layer\n",
    "    cells_fw = []\n",
    "    cells_bw = []\n",
    "    for _ in range(encoder_params['nb_layers']):\n",
    "        cell_fw, cell_bw = get_rnn_cell(encoder_params, is_training=is_training,\n",
    "                                        layer_name=layer_name)\n",
    "        cells_fw.append(cell_fw)\n",
    "        cells_bw.append(cell_bw)\n",
    "\n",
    "    if encoder_params['is_bidirectional']:\n",
    "        # outputs: concatenated fw and bw hidden states of last layer\n",
    "        # output_state_fw, output_state_bw = final_state\n",
    "        # output_states_fw is the final states, one tensor per layer,\n",
    "        # of the forward rnn.\n",
    "        # output_states_bw is the final states, one tensor per layer,\n",
    "        # of the backward rnn.\n",
    "\n",
    "        outputs, output_states_fw, output_states_bw, _ = \\\n",
    "            stack_bidirectional_dynamic_rnn(\n",
    "                cells_fw,\n",
    "                cells_bw,\n",
    "                inputs=layer_inputs,\n",
    "                initial_states_fw=None,\n",
    "                initial_states_bw=None,\n",
    "                dtype=tf.float32,\n",
    "                sequence_length=sequence_lengths,\n",
    "                parallel_iterations=None,\n",
    "                scope=layer_name)\n",
    "        final_state = (output_states_fw, output_states_bw)\n",
    "    else:\n",
    "        multi_rnn_cell = MultiRNNCell(cells_fw, state_is_tuple=True)\n",
    "        outputs, final_state = tf.nn.dynamic_rnn(\n",
    "            cell=multi_rnn_cell, inputs=layer_inputs,\n",
    "            sequence_length=sequence_lengths, initial_state=None,\n",
    "            dtype=tf.float32, parallel_iterations=None,\n",
    "            time_major=False, scope=layer_name)\n",
    "\n",
    "    return EncoderOutput(\n",
    "        outputs=outputs,\n",
    "        final_state=final_state)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "def _inference(input_sequences, is_training,\n",
    "               sequence_lengths=None):\n",
    "    \"\"\"\n",
    "\n",
    "    :param input_sequences:\n",
    "    :param is_training:\n",
    "    :param sequence_lengths:\n",
    "    :return:\n",
    "    \"\"\"\n",
    "\n",
    "    input_features = input_sequences.features\n",
    "    frame_sequence_lengths = sequence_lengths.frame_sequence_lengths\n",
    "\n",
    "    # Encoder\n",
    "    encoder_params = {\n",
    "        \"feat_dim\": params['feat_dim'],\n",
    "        \"nb_classes\": params['encoder_nb_classes'],\n",
    "        \"max_nb_frames\": params['max_nb_frames'],\n",
    "        # Model parameters\n",
    "        \"nb_hidden_units\": params['encoder_nb_hidden_units'],\n",
    "        \"cell_type\": params['encoder_cell_type'],\n",
    "        \"is_bidirectional\": params['encoder_is_bidirectional'],\n",
    "        \"activation\": params['encoder_activation'],\n",
    "        \"initializer\": params['encoder_initializer'],\n",
    "        \"init_scale\": params['encoder_init_scale'],\n",
    "        \"clip_gradients\": params['encoder_clip_gradients'],\n",
    "        \"nb_layers\": params['encoder_nb_layers'],\n",
    "        \"seed\": params['seed'],\n",
    "        \"log_dir\": params['log_dir'],\n",
    "        \"dropout_rate\": params['dropout_rate']\n",
    "    }\n",
    "\n",
    "    encoder_output = rnn_encoder(\n",
    "        input_sequences=input_features, is_training=is_training,\n",
    "        sequence_lengths=frame_sequence_lengths,\n",
    "        encoder_params=encoder_params)\n",
    "\n",
    "    # Encoder Time Distributed Fully Connected Layer\n",
    "    # for segmentation task\n",
    "    frame_logits, frame_predictions, frame_y_pred = output_fc_layer(\n",
    "        encoder_output.outputs, initializer=None)\n",
    "\n",
    "    return InferenceOutput(\n",
    "        frame_logits=frame_logits,\n",
    "        frame_predictions=frame_predictions,\n",
    "        frame_y_pred=frame_y_pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4c) Now we will define our loss and training operation (These ops are only used during training)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "def _loss(logits, labels, sample_weights=None):\n",
    "    \"\"\"\n",
    "    Args:\n",
    "        logits: encoder output logits\n",
    "                (3D Tensor [batch_size, max_len, nb_classes])\n",
    "        labels: ground truth frame level labels in one-hot encoding\n",
    "                [batch_size, max_len, nb_classes]\n",
    "        sample_weights: 2D Tensor [batch_size, max_len] with\n",
    "                        0 for all timesteps that are padded.\n",
    "    Returns:\n",
    "        loss_op: masked cross-entropy loss op\n",
    "    \"\"\"\n",
    "    # [batch_size, max_nb_frames, nb_classes]\n",
    "    frame_logits = logits.frame_logits\n",
    "    # [batch_size, max_nb_frames, nb_classes]\n",
    "    frame_labels = labels.frame_labels\n",
    "    # [batch_size, max_nb_frames]\n",
    "    frame_sample_weights = sample_weights.frame_sample_weights\n",
    "    \n",
    "    with tf.name_scope(\"cross_entropy_sequence_loss\"):\n",
    "        # Reshape logits, labels, sample_weights\n",
    "        # [batch_size, max_len, nb_classes] ->\n",
    "        # [batch_size*max_len, nb_classes]\n",
    "        tf.Print(frame_logits, [tf.shape(frame_logits)], \"Logits shape in loss\")\n",
    "        tf.Print(frame_labels, [tf.shape(frame_labels)], \"Labels shape in loss\")\n",
    "        nb_classes = array_ops.shape(frame_logits)[2]\n",
    "        flat_frame_logits = array_ops.reshape(frame_logits, [-1, nb_classes])\n",
    "        flat_frame_labels = array_ops.reshape(frame_labels, [-1, nb_classes])\n",
    "        # [batch_size, max_len] -> [batch_size*max_len]\n",
    "        # Compute cross-entropy for each frame separately\n",
    "        # flat_xent: 1D Tensor (batch_size*max_len,)\n",
    "        flat_frame_sample_weights = array_ops.reshape(\n",
    "            frame_sample_weights, [-1])\n",
    "        flat_frame_xent = nn_ops.softmax_cross_entropy_with_logits(\n",
    "            labels=flat_frame_labels, logits=flat_frame_logits)\n",
    "        # Apply masking on cent, setting to zero summands of\n",
    "        # the cross-entropy which correspond to padded frames\n",
    "        weighted_frame_xent = tf.multiply(flat_frame_xent, \n",
    "                                          flat_frame_sample_weights)\n",
    "        # Compute average cross-entropy loss over batches and non-padded\n",
    "        # timesteps\n",
    "        frame_xent = math_ops.reduce_sum(weighted_frame_xent)\n",
    "        nb_unpadded_frames = math_ops.reduce_sum(flat_frame_sample_weights)\n",
    "        nb_unpadded_frames += 1e-12  # avoid division by 0 for all-0 weights\n",
    "        frame_xent /= nb_unpadded_frames\n",
    "        frame_loss_op = frame_xent\n",
    "        tf.summary.scalar('frame_loss', frame_loss_op)\n",
    "\n",
    "    return frame_loss_op"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "def _grad_vars():\n",
    "    g_vars = tf.trainable_variables()\n",
    "    return g_vars\n",
    "\n",
    "\n",
    "def _train_op(loss):\n",
    "    # Add a scalar summary for loss.\n",
    "    tf.summary.scalar('loss', loss)\n",
    "\n",
    "    # Create a variable to track the global step.\n",
    "    global_step = tf.Variable(0, name='global_step', trainable=False)\n",
    "\n",
    "    # Learning rate scheduler\n",
    "    if params['decay_rate'] > 0:\n",
    "        learning_rate = tf.train.exponential_decay(\n",
    "            learning_rate=params['learning_rate'],\n",
    "            global_step=global_step,\n",
    "            decay_steps=params['decay_period'],\n",
    "            decay_rate=params['decay_rate'], staircase=True)\n",
    "    else:\n",
    "        learning_rate = params['learning_rate']\n",
    "    # Create optimizer\n",
    "    optimizer_name = params['optimizer_name']\n",
    "    if optimizer_name == 'sgd':\n",
    "        optimizer = tf.train.MomentumOptimizer(\n",
    "            learning_rate=learning_rate,\n",
    "            momentum=params['momentum'],\n",
    "            use_nesterov=params['nesterov'])\n",
    "    elif optimizer_name == 'rmsprop':\n",
    "        optimizer = tf.train.RMSPropOptimizer(learning_rate=learning_rate)\n",
    "    elif optimizer_name == 'Adagrad':\n",
    "        optimizer = tf.train.AdagradOptimizer(learning_rate=learning_rate)\n",
    "    elif optimizer_name == 'Adadelta':\n",
    "        optimizer = tf.train.AdadeltaOptimizer(learning_rate=learning_rate)\n",
    "    elif optimizer_name == 'Adam':\n",
    "        optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate)\n",
    "    else:\n",
    "        raise ValueError(\"Not supported optimizer: %s\", optimizer_name)\n",
    "\n",
    "    # Use the optimizer to apply the gradients that minimize the loss\n",
    "    # and to increment the global step counter as a single training step.\n",
    "    # train_op = optimizer.minimize(loss, global_step=global_step,\n",
    "    # name=name)\n",
    "    g_vars = _grad_vars()\n",
    "\n",
    "    # Compute gradients\n",
    "    grads_and_vars = optimizer.compute_gradients(\n",
    "        loss, var_list=g_vars)\n",
    "\n",
    "    vars_with_grad = [v for g, v in grads_and_vars if g is not None]\n",
    "    print(\"vars_with_grad: \", vars_with_grad)\n",
    "    if not vars_with_grad:\n",
    "        raise ValueError(\n",
    "            \"No gradients provided for any variable, \"\n",
    "            \"check your graph for ops \"\n",
    "            \"that do not support gradients, \"\n",
    "            \"between variables %s and loss %s.\" %\n",
    "            ([str(v) for _, v in grads_and_vars], loss))\n",
    "\n",
    "    # Optionally clip gradients by global norm.\n",
    "    if params['encoder_clip_gradients'] is not None:\n",
    "        clipped_grads_and_vars = clip_gradients(\n",
    "            grads_and_vars, params['encoder_clip_gradients'])\n",
    "    else:\n",
    "        clipped_grads_and_vars = grads_and_vars\n",
    "\n",
    "    # Add histograms for variables, gradients and gradient norms\n",
    "    for gradient, variable in clipped_grads_and_vars:\n",
    "        if isinstance(gradient, ops.IndexedSlices):\n",
    "            grad_values = gradient.values\n",
    "        else:\n",
    "            grad_values = gradient\n",
    "\n",
    "        if grad_values is not None:\n",
    "            var_name = variable.name.replace(\":\", \"_\")\n",
    "            tf.summary.histogram(\"gradients/%s\" % var_name,\n",
    "                                 grad_values)\n",
    "            tf.summary.scalar(\"gradient_norm/%s\" % var_name,\n",
    "                              clip_ops.global_norm([grad_values]))\n",
    "\n",
    "    if params['encoder_clip_gradients'] is not None:\n",
    "        for gradient, variable in grads_and_vars:\n",
    "            if isinstance(gradient, ops.IndexedSlices):\n",
    "                grad_values = gradient.values\n",
    "            else:\n",
    "                grad_values = gradient\n",
    "\n",
    "            if grad_values is not None:\n",
    "                var_name = variable.name.replace(\":\", \"_\")\n",
    "                tf.summary.histogram(\"gradients_before_clip/%s\" % var_name,\n",
    "                                     grad_values)\n",
    "                tf.summary.scalar(\"gradient_norm_before_clip/%s\" % var_name,\n",
    "                                  clip_ops.global_norm([grad_values]))\n",
    "\n",
    "    # Create gradient updates.\n",
    "    train_op = optimizer.apply_gradients(clipped_grads_and_vars,\n",
    "                                         global_step=global_step,\n",
    "                                         name=\"train\")\n",
    "    return train_op"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4d) Finally, we declare placeholders for feeding data to our graph during training/testing."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "def _placeholder_inputs():\n",
    "    \"\"\"Generate placeholder variables to represent the input tensors.\n",
    "    These placeholders are used as inputs for the model.\n",
    "    Returns: dictionary of placeholders with keys/values:\n",
    "      'features_pl': Sequences placeholder.\n",
    "      'frame_labels_pl': Frame labels placeholder.\n",
    "      'frame_sample_weights_pl': Frame sample weight placeholder\n",
    "      'frame_sequence_lengths_pl': Frame sequence lengths placeholder\n",
    "    \"\"\"\n",
    "\n",
    "    features_placeholder = tf.placeholder(\n",
    "        tf.float32,\n",
    "        [None, params['max_nb_frames'], params['feat_dim']],\n",
    "        name=\"features_pl\")\n",
    "\n",
    "    frame_labels_placeholder = tf.placeholder(\n",
    "        tf.int32,\n",
    "        [None, params['max_nb_frames'],\n",
    "         params['encoder_nb_classes']],\n",
    "        name=\"frame_labels_pl\")\n",
    "    \n",
    "    frame_sample_weights = tf.placeholder(\n",
    "        tf.float32,\n",
    "        [None, params['max_nb_frames']],\n",
    "        name=\"frame_sample_weights_pl\")\n",
    "    \n",
    "    frame_sequence_lengths_placeholder = tf.placeholder(\n",
    "        tf.int32,\n",
    "        [None],\n",
    "        name=\"frame_sequence_lengths_pl\")\n",
    "\n",
    "    is_training_placeholder = tf.placeholder(\n",
    "        tf.bool, [],\n",
    "        name=\"is_training_pl\")\n",
    "\n",
    "    placeholders = {\n",
    "        'features_pl': features_placeholder,\n",
    "        'frame_labels_pl': frame_labels_placeholder,\n",
    "        'frame_sample_weights_pl': frame_sample_weights,\n",
    "        'frame_sequence_lengths_pl':\n",
    "            frame_sequence_lengths_placeholder,\n",
    "        'is_training_pl':\n",
    "            is_training_placeholder,\n",
    "    }\n",
    "    return placeholders\n",
    "\n",
    "\n",
    "def _fill_feed_dict(generator, placeholders_dict, is_training):\n",
    "    \"\"\"Fills the feed_dict for training the given step.\n",
    "    A feed_dict takes the form of:\n",
    "    feed_dict = {\n",
    "        <placeholder>: <tensor of values to be passed for placeholder>,\n",
    "        ....\n",
    "    }\n",
    "    Args:\n",
    "        generator: instance of SequenceBatchGenerator class\n",
    "                   use methods .__next__() for getting next batch from\n",
    "                   data and .get_all_samples() for getting\n",
    "                   all samples\n",
    "        placeholders_dict: dictionary with keys corresponding to placeholder\n",
    "         names and values consisting of the actual placeholders\n",
    "    Returns:\n",
    "      feed_dict: The feed dictionary mapping from placeholders to values.\n",
    "      values_dict: The same dictionary but with strings as keys and\n",
    "                   not placeholders\n",
    "    \"\"\"\n",
    "\n",
    "    # Create the feed_dict for the placeholders filled with the next\n",
    "    # `batch size` examples.\n",
    "    features, frame_labels, frame_sample_weights, \\\n",
    "        frame_sequence_lengths = \\\n",
    "        generator.__next__()\n",
    "    feed_dict = {\n",
    "        placeholders_dict['features_pl']: features,\n",
    "        placeholders_dict['frame_labels_pl']: frame_labels,\n",
    "        placeholders_dict[\n",
    "            'frame_sample_weights_pl']: frame_sample_weights,\n",
    "        placeholders_dict[\n",
    "            'frame_sequence_lengths_pl']: frame_sequence_lengths,\n",
    "        placeholders_dict['is_training_pl']: is_training,\n",
    "    }\n",
    "    values_dict = {\n",
    "        'features': features,\n",
    "        'frame_labels': frame_labels,\n",
    "        'frame_sample_weights': frame_sample_weights,\n",
    "        'frame_sequence_lengths': frame_sequence_lengths,\n",
    "        'is_training': is_training,\n",
    "    }\n",
    "\n",
    "\n",
    "    return feed_dict, values_dict"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 5) Functions for launching graph and training model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Building, launching and executing operations on the graph\n",
    "def _do_train_eval(sess, train_op, loss, eval_metric, predictions,\n",
    "                   y_pred, training_generator, validation_generator,\n",
    "                   placeholders_dict, summary, summary_writer, logits):\n",
    "    \"\"\"Performs training\n",
    "    Args:\n",
    "      sess: The session in which the model has been initialized\n",
    "      train_op: Training operation\n",
    "      loss: namedtuple (\"frame_loss\")\n",
    "      eval_metric: namedtuple (\"frame_eval_metric\")\n",
    "      predictions: namedtuple (\"frame_predictions\")\n",
    "      y_pred: namedtuple (\"frame_y_pred\")\n",
    "      training_generator: instance of class SequenceBatchGenerator\n",
    "      validation_generator: instance of class SequenceBatchGenerator\n",
    "      placeholders_dict: dictionary of placeholders\n",
    "      summary_writer, summary\n",
    "      logits: namedtuple (\"frame_logits\")\n",
    "    Returns:\n",
    "        avg_metric: average eval metric evaluated over all validation\n",
    "                    sequences (\"frame_avg_metric\")\n",
    "        train_loss: average validation loss over all validation sequences\n",
    "                  (\"train_frame_loss\")\n",
    "        predictions: 3D Tensor [nb_train_samples, max_len, nb_classes]\n",
    "                  (\"frame_predictions\")\n",
    "        logits: 3D Tensor [nb_train_samples, max_len, nb_classes]\n",
    "    \"\"\"\n",
    "\n",
    "    frame_loss = loss.frame_loss\n",
    "    frame_eval_metric = eval_metric.frame_eval_metric\n",
    "    frame_predictions = predictions.frame_predictions\n",
    "    frame_logits = logits.frame_logits\n",
    "    frame_y_pred = y_pred.frame_y_pred\n",
    "\n",
    "    # Start the training loop.\n",
    "    nb_samples = training_generator.__len__()\n",
    "    nb_steps = training_generator.steps\n",
    "    logging.info('Nb of mini-batches per epoch: %d', nb_steps)\n",
    "\n",
    "    # Train\n",
    "    epoch = 0\n",
    "    avg_frame_metric = 0\n",
    "    train_frame_loss = 0\n",
    "    train_frame_loss_sum = 0\n",
    "    train_frame_y_true = []\n",
    "    train_frame_y_pred = []\n",
    "    train_frame_sample_weights = []\n",
    "\n",
    "    val_frame_res = None\n",
    "\n",
    "    start_time = time.time()\n",
    "\n",
    "    optimization_log = defaultdict(list)\n",
    "    for iteration in range(params['nb_epochs'] * nb_steps):\n",
    "        if iteration % nb_steps == 0:\n",
    "            epoch += 1\n",
    "\n",
    "            train_frame_loss_sum = 0\n",
    "            train_frame_y_true = []\n",
    "            train_frame_y_pred = []\n",
    "            train_frame_sample_weights = []\n",
    "\n",
    "            start_time = time.time()\n",
    "\n",
    "        # Fill a feed dictionary with current batch of\n",
    "        # training this particular training step.\n",
    "        training_feed_dict, training_values_dict = _fill_feed_dict(\n",
    "            training_generator, placeholders_dict, is_training=1)\n",
    "\n",
    "        # Run one step of the model.  The return values are the\n",
    "        # activations from the `train_op` (which is discarded)\n",
    "        # and the `loss` Op.  To inspect the values of your Ops or\n",
    "        # variables, you may include them\n",
    "        # in the list passed to sess.run() and the value tensors will be\n",
    "        # returned in the tuple from the call.\n",
    "\n",
    "        # Train with one batch and evaluate\n",
    "        sess_run_ops = [train_op, frame_loss,\n",
    "                        frame_predictions, frame_logits, frame_y_pred]\n",
    "        sess_run_res = sess.run(sess_run_ops, feed_dict=training_feed_dict)\n",
    "        train_op_res, train_frame_loss_batch, \\\n",
    "            train_frame_predictions_batch, train_frame_logits_batch, \\\n",
    "            train_frame_y_pred_batch, \\\n",
    "            = sess_run_res\n",
    "\n",
    "        train_frame_loss_sum += train_frame_loss_batch\n",
    "\n",
    "        train_frame_y_pred.append(train_frame_y_pred_batch)\n",
    "        train_frame_y_true.append(argmax(\n",
    "            training_values_dict['frame_labels'], axis=-1))\n",
    "        train_frame_sample_weights.append(\n",
    "            training_values_dict['frame_sample_weights'])\n",
    "\n",
    "        # Write the summaries, print an overview and evaluate the model\n",
    "        # at the end of every epoch\n",
    "        if (iteration + 1) % nb_steps == 0:\n",
    "            duration = time.time() - start_time\n",
    "            # Results aggregation over all batches\n",
    "            train_frame_loss = float(train_frame_loss_sum) / nb_steps\n",
    "\n",
    "            # train_frame_y_pred: (nb_samples, nb_timesteps, nb_classes)\n",
    "            train_frame_y_pred = vstack(train_frame_y_pred)\n",
    "            # train_frame_y_true: (nb_samples, nb_timesteps, nb_classes)\n",
    "            train_frame_y_true = vstack(train_frame_y_true)\n",
    "            train_frame_sample_weights = vstack(train_frame_sample_weights)\n",
    "            avg_frame_metric = frame_eval_metric(\n",
    "                train_frame_y_true, train_frame_y_pred,\n",
    "                sample_weights=train_frame_sample_weights)\n",
    "\n",
    "            logging.info('Epoch %d: , '\n",
    "                         'nb_train_samples: %d, train_frame_metric: %f, '\n",
    "                         'train_frame_loss: %f, '\n",
    "                         'duration: %f',\n",
    "                         epoch,  nb_samples,\n",
    "                         avg_frame_metric,\n",
    "                         train_frame_loss,\n",
    "                         duration)\n",
    "\n",
    "            optimization_log['train_frame_loss'].append(train_frame_loss)\n",
    "            optimization_log['train_frame_metric'].append(avg_frame_metric)\n",
    "\n",
    "            # Update the events file.\n",
    "            summary_str = sess.run(summary,\n",
    "                                   feed_dict=training_feed_dict)\n",
    "            summary_writer.add_summary(summary_str, iteration)\n",
    "            summary_writer.flush()\n",
    "\n",
    "            val_frame_res = _do_eval(\n",
    "                sess=sess, loss=loss, eval_metric=eval_metric,\n",
    "                predictions=predictions, y_pred=y_pred,\n",
    "                generator=validation_generator,\n",
    "                placeholders_dict=placeholders_dict,\n",
    "                logits=logits)\n",
    "\n",
    "            optimization_log['val_frame_loss'].append(\n",
    "                val_frame_res.val_frame_loss)\n",
    "            optimization_log['val_frame_metric'].append(\n",
    "                val_frame_res.val_frame_metric)\n",
    "\n",
    "            # TODO: Log output using output_logger if available\n",
    "\n",
    "    train_frame_res = TrainFrameResults(\n",
    "        train_frame_metric=avg_frame_metric,\n",
    "        train_frame_loss=train_frame_loss,\n",
    "        train_frame_predictions=None,\n",
    "        train_frame_y_pred=train_frame_y_pred,\n",
    "        train_frame_y_true=train_frame_y_true,\n",
    "        train_frame_sample_weights=train_frame_sample_weights,\n",
    "        train_frame_logits=None,\n",
    "    )\n",
    "\n",
    "    plot_optimization_log_frame(optimization_log,\n",
    "                                params['log_dir'])\n",
    "    save_to_pickle(os.path.join(params['log_dir'], 'optimization_log.p'),\n",
    "                   optimization_log)\n",
    "\n",
    "    return train_frame_res, val_frame_res\n",
    "\n",
    "\n",
    "def _do_eval(sess, loss, eval_metric, predictions, y_pred,\n",
    "             generator, placeholders_dict, logits):\n",
    "    \"\"\"Runs one evaluation against all data samples, batch by batch.\n",
    "    Args:\n",
    "      sess: The session in which the model has been trained.\n",
    "      loss: namedtuple (\"frame_loss\")\n",
    "      eval_metric: namedtuple (\"frame_eval_metric\")\n",
    "      predictions: namedtuple (\"frame_predictions\")\n",
    "      y_pred: namedtuple (\"frame_y_pred\")\n",
    "      generator: instance of class SequenceBatchGenerator\n",
    "      placeholders_dict: dictionary of placeholders\n",
    "      logits: namedtuple (\"frame_logits\")\n",
    "    Returns:\n",
    "        avg_metric: average eval metric evaluated over all validation\n",
    "                    sequences (\"frame_avg_metric\")\n",
    "        val_loss: average validation loss over all validation sequences\n",
    "                  (\"val_frame_loss\")\n",
    "        predictions: 3D Tensor [nb_val_samples, max_len, nb_classes]\n",
    "                  (\"frame_predictions\")\n",
    "        y_pred_val: 3D Tensor [nb_val_samples, max_len]\n",
    "                  (\"frame_y_pred_val\")\n",
    "        y_true_val: 3D Tensor [nb_val_samples, max_len]\n",
    "                  (\"frame_y_true_val\")\n",
    "    \"\"\"\n",
    "\n",
    "    frame_loss = loss.frame_loss\n",
    "    frame_eval_metric = eval_metric.frame_eval_metric\n",
    "    frame_predictions = predictions.frame_predictions\n",
    "    frame_logits = logits.frame_logits\n",
    "    frame_y_pred = y_pred.frame_y_pred\n",
    "\n",
    "    # nb_samples: number of samples to evaluate against\n",
    "    # nb_steps: number of steps until we see all nb_samples\n",
    "    nb_samples = generator.__len__()\n",
    "    nb_steps = generator.steps\n",
    "\n",
    "    # And run one epoch of eval.\n",
    "\n",
    "    val_frame_loss_sum = 0\n",
    "    val_frame_predictions = []\n",
    "    val_frame_logits = []\n",
    "    val_frame_y_pred = []\n",
    "    val_frame_y_true = []\n",
    "    val_frame_sample_weights = []\n",
    "\n",
    "    for step in range(nb_steps):\n",
    "\n",
    "        feed_dict, values_dict = _fill_feed_dict(\n",
    "            generator, placeholders_dict, is_training=0)\n",
    "\n",
    "        # Frame segmentation evaluation\n",
    "        val_frame_loss_batch, \\\n",
    "            val_frame_predictions_batch, val_frame_logits_batch, \\\n",
    "            val_frame_y_pred_batch = sess.run(\n",
    "                [frame_loss, frame_predictions, frame_logits, frame_y_pred],\n",
    "                feed_dict=feed_dict)\n",
    "\n",
    "        val_frame_loss_sum += val_frame_loss_batch\n",
    "        val_frame_predictions.append(array(val_frame_predictions_batch))\n",
    "        val_frame_logits.append(array(val_frame_logits_batch))\n",
    "        val_frame_y_pred.append(val_frame_y_pred_batch)\n",
    "        val_frame_y_true.append(argmax(values_dict['frame_labels'], axis=-1))\n",
    "        val_frame_sample_weights.append(values_dict['frame_sample_weights'])\n",
    "\n",
    "    # Results aggregation over all batches\n",
    "    val_frame_loss = float(val_frame_loss_sum) / nb_steps\n",
    "    val_frame_predictions = vstack(val_frame_predictions)\n",
    "    val_frame_logits = vstack(val_frame_logits)\n",
    "    val_frame_y_pred = vstack(val_frame_y_pred)\n",
    "    val_frame_y_true = vstack(val_frame_y_true)\n",
    "    val_frame_sample_weights = vstack(val_frame_sample_weights)\n",
    "\n",
    "    avg_frame_metric = frame_eval_metric(\n",
    "        val_frame_y_true, val_frame_y_pred,\n",
    "        sample_weights=val_frame_sample_weights)\n",
    "\n",
    "    logging.info('nb_val_samples: %d, val_frame_metric: %f, '\n",
    "                 'val_frame_loss: %f',\n",
    "                 nb_samples, avg_frame_metric,\n",
    "                 val_frame_loss)\n",
    "\n",
    "    frame_res = ValFrameResults(\n",
    "        val_frame_metric=avg_frame_metric,\n",
    "        val_frame_loss=val_frame_loss,\n",
    "        val_frame_predictions=val_frame_predictions,\n",
    "        val_frame_y_pred=val_frame_y_pred,\n",
    "        val_frame_y_true=val_frame_y_true,\n",
    "        val_frame_sample_weights=val_frame_sample_weights,\n",
    "        val_frame_logits=val_frame_logits,\n",
    "    )\n",
    "\n",
    "    return frame_res"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "def fit_generator(training_generator, validation_generator=None):\n",
    "    \"\"\"Fit SupSeq2SeqTF Model to training_data and validate\n",
    "    on validation data\n",
    "    Args:\n",
    "        training_generator: training data generator.\n",
    "        instance of class SequenceBatchGenerator,\n",
    "        instance of class SequenceBatchGenerator,\n",
    "         or any class implementing methods: __next__() getting next batch\n",
    "          (X, y, sample_weights, sequence_lengths), get_all_samples()\n",
    "          getting all samples (X, y, sample_weights, sequence_lengths)\n",
    "          , __len__(): get number of samples, steps: get number of steps\n",
    "          required to get all samples if we load them in a batch fashion\n",
    "          and batch_size: batch size\n",
    "        validation_generator: validation data generator\n",
    "    \"\"\"\n",
    "\n",
    "    nb_steps = training_generator.steps\n",
    "    # Convert decay_rate from epochs to iterations\n",
    "    params['decay_period'] *= nb_steps\n",
    "\n",
    "    # Tell TensorFlow that the model will be built into the default Graph.\n",
    "    with tf.Graph().as_default():\n",
    "        tf.set_random_seed(params['seed'])\n",
    "\n",
    "        # Generate placeholders for training/validation data\n",
    "        placeholders_dict = _placeholder_inputs()\n",
    "\n",
    "        # Build a Graph that computes predictions from the inference model.\n",
    "        input_sequences_tuple = namedtuple(\"InputSequences\",\n",
    "                                           \"features\")\n",
    "        input_sequences = input_sequences_tuple(\n",
    "            features=placeholders_dict['features_pl']\n",
    "        )\n",
    "        sequence_lengths_tuple = namedtuple(\n",
    "            \"SequenceLengths\",\n",
    "            \"frame_sequence_lengths\")\n",
    "        sequence_lengths = sequence_lengths_tuple(\n",
    "            frame_sequence_lengths=placeholders_dict[\n",
    "                'frame_sequence_lengths_pl'],\n",
    "        )\n",
    "\n",
    "        inference_output = _inference(\n",
    "            input_sequences=input_sequences,\n",
    "            is_training=placeholders_dict['is_training_pl'],\n",
    "            sequence_lengths=sequence_lengths)\n",
    "\n",
    "        # Parse inference output\n",
    "        frame_logits = inference_output.frame_logits\n",
    "        frame_predictions = inference_output.frame_predictions\n",
    "        frame_y_pred = inference_output.frame_y_pred\n",
    "\n",
    "        # Create inputs for _loss and _do_train_eval functions\n",
    "        logits_tuple = namedtuple(\"logits\",\n",
    "                                  \"frame_logits\")\n",
    "        logits = logits_tuple(\n",
    "            frame_logits=frame_logits,\n",
    "        )\n",
    "        labels_tuple = namedtuple(\"labels\",\n",
    "                                  \"frame_labels\")\n",
    "        labels = labels_tuple(\n",
    "            frame_labels=placeholders_dict['frame_labels_pl'],\n",
    "        )\n",
    "        sample_weights_tuple = namedtuple(\n",
    "            \"sample_weights\",\n",
    "            \"frame_sample_weights\")\n",
    "        sample_weights = sample_weights_tuple(\n",
    "            frame_sample_weights=placeholders_dict[\n",
    "                'frame_sample_weights_pl'],\n",
    "        )\n",
    "\n",
    "        # Add to the Graph the Ops for loss calculation.\n",
    "        frame_loss = _loss(\n",
    "            logits=logits, labels=labels,\n",
    "            sample_weights=sample_weights\n",
    "        )\n",
    "\n",
    "        grad_vars = _grad_vars()\n",
    "        # Add to the Graph the Ops that calculate and apply gradients.\n",
    "        train_op = _train_op(frame_loss)\n",
    "\n",
    "        # Add the Op to compare predictions to labels during evaluation.\n",
    "        loss_tuple = namedtuple(\"loss\",\n",
    "                                \"frame_loss\")\n",
    "        loss = loss_tuple(\n",
    "            frame_loss=frame_loss,\n",
    "        )\n",
    "        eval_metric_tuple = namedtuple(\"eval_metric\",\n",
    "                                       \"frame_eval_metric\")\n",
    "        eval_metric = eval_metric_tuple(\n",
    "            frame_eval_metric=per_frame_accuracy,\n",
    "        )\n",
    "        predictions_tuple = namedtuple(\"predictions\",\n",
    "                                       \"frame_predictions\")\n",
    "        predictions = predictions_tuple(\n",
    "            frame_predictions=frame_predictions,\n",
    "        )\n",
    "        y_pred_tuple = namedtuple(\"y_pred\",\n",
    "                                  \"frame_y_pred\")\n",
    "        y_pred = y_pred_tuple(\n",
    "            frame_y_pred=frame_y_pred,\n",
    "        )\n",
    "\n",
    "        # Build the summary Tensor based on the TF collection of Summaries.\n",
    "        summary = tf.summary.merge_all()\n",
    "\n",
    "        # Add the variable initializer Op.\n",
    "        init = tf.global_variables_initializer()\n",
    "        init_l = tf.local_variables_initializer()\n",
    "\n",
    "        # Create a saver for writing training checkpoints.\n",
    "        saver = tf.train.Saver()\n",
    "\n",
    "        # Create a session for running Ops on the Graph.\n",
    "        sess = tf.Session()\n",
    "\n",
    "        # Instantiate a SummaryWriter to output summaries and the Graph.\n",
    "        summary_writer = tf.summary.FileWriter(params['log_dir'],\n",
    "                                               sess.graph)\n",
    "\n",
    "        # And then after everything is built:\n",
    "        # Run the Op to initialize the variables.\n",
    "        sess.run(init)\n",
    "        sess.run(init_l)\n",
    "\n",
    "        # Print number of trainable parameters\n",
    "        nb_trainable_params = sum([prod(v.get_shape().as_list())\n",
    "                                   for v in grad_vars])\n",
    "        logging.info(\"Number of trainable params: %d\", nb_trainable_params)\n",
    "        # Train and evaluate at each epoch\n",
    "        nb_training_steps_per_epoch = training_generator.steps\n",
    "        train_frame_res, val_frame_res = \\\n",
    "            _do_train_eval(\n",
    "                sess=sess, train_op=train_op, loss=loss,\n",
    "                eval_metric=eval_metric, predictions=predictions,\n",
    "                y_pred=y_pred,\n",
    "                training_generator=training_generator,\n",
    "                validation_generator=validation_generator,\n",
    "                placeholders_dict=placeholders_dict,\n",
    "                summary=summary, summary_writer=summary_writer,\n",
    "                logits=logits)\n",
    "\n",
    "        # Save checkpoint\n",
    "        checkpoint_file = os.path.join(params['log_dir'],\n",
    "                                       'multilabel_rnn_model.ckpt')\n",
    "        nb_total_steps = params[\n",
    "                             'nb_epochs'] * nb_training_steps_per_epoch\n",
    "        saver.save(sess, checkpoint_file,\n",
    "                   global_step=nb_total_steps - 1)\n",
    "\n",
    "        # TODO: Log output using output_logger if available\n",
    "\n",
    "        return val_frame_res, train_frame_res"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "def restore_and_eval(checkpoint_file, training_generator,\n",
    "        validation_generator=None):\n",
    "    \"\"\"Restores or uses randomly initialized SupSeq2SeqTF Model\n",
    "       and evaluates it on validation data\n",
    "    Args:\n",
    "        training_generator: training data generator.\n",
    "        instance of class SequenceBatchGenerator,\n",
    "         or any class implementing methods: __next__() getting next batch\n",
    "          (X, y, sample_weights, sequence_lengths), get_all_samples()\n",
    "          getting all samples (X, y, sample_weights, sequence_lengths)\n",
    "          , __len__(): get number of samples, steps: get number of steps\n",
    "          required to get all samples if we load them in a batch fashion\n",
    "          and batch_size: batch size\n",
    "        validation_generator: validation data generator\n",
    "        params:\n",
    "    \"\"\"\n",
    "\n",
    "    nb_steps = training_generator.steps\n",
    "    # Convert decay_rate from epochs to iterations\n",
    "    # TODO: refactor\n",
    "    params['decay_period'] *= nb_steps\n",
    "\n",
    "    # Tell TensorFlow that the model will be built into the default Graph.\n",
    "    with tf.Graph().as_default():\n",
    "        tf.set_random_seed(params['seed'])\n",
    "\n",
    "        # Generate placeholders for training/validation data\n",
    "        placeholders_dict = _placeholder_inputs()\n",
    "\n",
    "        # Build a Graph that computes predictions from the inference model.\n",
    "        input_sequences_tuple = namedtuple(\"InputSequences\",\n",
    "                                           \"features\")\n",
    "        input_sequences = input_sequences_tuple(\n",
    "            features=placeholders_dict['features_pl']\n",
    "        )\n",
    "        sequence_lengths_tuple = namedtuple(\n",
    "            \"SequenceLengths\",\n",
    "            \"frame_sequence_lengths\")\n",
    "        sequence_lengths = sequence_lengths_tuple(\n",
    "            frame_sequence_lengths=placeholders_dict[\n",
    "                'frame_sequence_lengths_pl'],\n",
    "        )\n",
    "\n",
    "        inference_output = _inference(\n",
    "            input_sequences=input_sequences,\n",
    "            is_training=placeholders_dict['is_training_pl'],\n",
    "            sequence_lengths=sequence_lengths)\n",
    "\n",
    "        # Parse inference output\n",
    "        frame_logits = inference_output.frame_logits\n",
    "        frame_predictions = inference_output.frame_predictions\n",
    "        frame_y_pred = inference_output.frame_y_pred\n",
    "\n",
    "        # Create inputs for _loss and _do_train_eval functions\n",
    "        logits_tuple = namedtuple(\"logits\",\n",
    "                                  \"frame_logits\")\n",
    "        logits = logits_tuple(\n",
    "            frame_logits=frame_logits,\n",
    "        )\n",
    "        labels_tuple = namedtuple(\"labels\",\n",
    "                                  \"frame_labels\")\n",
    "        labels = labels_tuple(\n",
    "            frame_labels=placeholders_dict['frame_labels_pl'],\n",
    "        )\n",
    "        sample_weights_tuple = namedtuple(\n",
    "            \"sample_weights\",\n",
    "            \"frame_sample_weights\")\n",
    "        sample_weights = sample_weights_tuple(\n",
    "            frame_sample_weights=placeholders_dict[\n",
    "                'frame_sample_weights_pl'],\n",
    "        )\n",
    "\n",
    "        # Add to the Graph the Ops for loss calculation.\n",
    "        frame_loss = _loss(\n",
    "            logits=logits, labels=labels,\n",
    "            sample_weights=sample_weights\n",
    "        )\n",
    "\n",
    "        grad_vars = _grad_vars()\n",
    "        # Add to the Graph the Ops that calculate and apply gradients.\n",
    "        train_op = _train_op(frame_loss)\n",
    "\n",
    "        # Add the Op to compare predictions to labels during evaluation.\n",
    "        loss_tuple = namedtuple(\"loss\",\n",
    "                                \"frame_loss\")\n",
    "        loss = loss_tuple(\n",
    "            frame_loss=frame_loss,\n",
    "        )\n",
    "        eval_metric_tuple = namedtuple(\"eval_metric\",\n",
    "                                       \"frame_eval_metric\")\n",
    "        eval_metric = eval_metric_tuple(\n",
    "            frame_eval_metric=per_frame_accuracy,\n",
    "        )\n",
    "        predictions_tuple = namedtuple(\"predictions\",\n",
    "                                       \"frame_predictions\")\n",
    "        predictions = predictions_tuple(\n",
    "            frame_predictions=frame_predictions,\n",
    "        )\n",
    "        y_pred_tuple = namedtuple(\"y_pred\",\n",
    "                                  \"frame_y_pred\")\n",
    "        y_pred = y_pred_tuple(\n",
    "            frame_y_pred=frame_y_pred,\n",
    "        )\n",
    "\n",
    "        # Build the summary Tensor based on the TF collection of Summaries.\n",
    "        summary = tf.summary.merge_all()\n",
    "\n",
    "        # Create a saver for restoring training checkpoints.\n",
    "        saver = tf.train.Saver()\n",
    "\n",
    "        # Create a session for running Ops on the Graph.\n",
    "        sess = tf.Session()\n",
    "\n",
    "        # And then after everything is built:\n",
    "        # Restore variables from disk.\n",
    "        saver.restore(sess, checkpoint_file)\n",
    "        print(\"Model restored.\")\n",
    "\n",
    "        # Print number of trainable parameters\n",
    "        nb_trainable_params = sum([prod(v.get_shape().as_list())\n",
    "                                   for v in grad_vars])\n",
    "        logging.info(\"Number of trainable params: %d\", nb_trainable_params)\n",
    "        \n",
    "        # Evaluate at each epoch\n",
    "        val_frame_res = _do_eval(\n",
    "            sess=sess, loss=loss, eval_metric=eval_metric,\n",
    "            predictions=predictions, y_pred=y_pred,\n",
    "            generator=validation_generator,\n",
    "            placeholders_dict=placeholders_dict,\n",
    "            logits=logits)\n",
    "\n",
    "        return val_frame_res"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 6) Train and evaluate model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Layer inputs static shape:  (?, 650, 64)\n",
      "Logits static shape:  (?, 650, 48)\n",
      "WARNING:tensorflow:From <ipython-input-22-f9ef746222c3>:35: softmax_cross_entropy_with_logits (from tensorflow.python.ops.nn_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "\n",
      "Future major versions of TensorFlow will allow gradients to flow\n",
      "into the labels input on backprop by default.\n",
      "\n",
      "See tf.nn.softmax_cross_entropy_with_logits_v2.\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From <ipython-input-22-f9ef746222c3>:35: softmax_cross_entropy_with_logits (from tensorflow.python.ops.nn_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "\n",
      "Future major versions of TensorFlow will allow gradients to flow\n",
      "into the labels input on backprop by default.\n",
      "\n",
      "See tf.nn.softmax_cross_entropy_with_logits_v2.\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "vars_with_grad:  [<tf.Variable 'encoder_rnn/cell_0/bidirectional_rnn/fw/gru_cell/gates/kernel:0' shape=(320, 512) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_0/bidirectional_rnn/fw/gru_cell/gates/bias:0' shape=(512,) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_0/bidirectional_rnn/fw/gru_cell/candidate/kernel:0' shape=(320, 256) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_0/bidirectional_rnn/fw/gru_cell/candidate/bias:0' shape=(256,) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_0/bidirectional_rnn/bw/gru_cell/gates/kernel:0' shape=(320, 512) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_0/bidirectional_rnn/bw/gru_cell/gates/bias:0' shape=(512,) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_0/bidirectional_rnn/bw/gru_cell/candidate/kernel:0' shape=(320, 256) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_0/bidirectional_rnn/bw/gru_cell/candidate/bias:0' shape=(256,) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_1/bidirectional_rnn/fw/gru_cell/gates/kernel:0' shape=(768, 512) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_1/bidirectional_rnn/fw/gru_cell/gates/bias:0' shape=(512,) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_1/bidirectional_rnn/fw/gru_cell/candidate/kernel:0' shape=(768, 256) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_1/bidirectional_rnn/fw/gru_cell/candidate/bias:0' shape=(256,) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_1/bidirectional_rnn/bw/gru_cell/gates/kernel:0' shape=(768, 512) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_1/bidirectional_rnn/bw/gru_cell/gates/bias:0' shape=(512,) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_1/bidirectional_rnn/bw/gru_cell/candidate/kernel:0' shape=(768, 256) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_1/bidirectional_rnn/bw/gru_cell/candidate/bias:0' shape=(256,) dtype=float32_ref>, <tf.Variable 'encoder_time_fully_connected/kernel:0' shape=(512, 48) dtype=float32_ref>, <tf.Variable 'encoder_time_fully_connected/bias:0' shape=(48,) dtype=float32_ref>]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:Number of trainable params: 1698864\n",
      "INFO:root:Nb of mini-batches per epoch: 45\n",
      "INFO:root:Epoch 1: , nb_train_samples: 1423, train_frame_metric: 0.130866, train_frame_loss: 3.160713, duration: 63.395198\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.238291, val_frame_loss: 2.782112\n",
      "INFO:root:Epoch 2: , nb_train_samples: 1423, train_frame_metric: 0.249033, train_frame_loss: 2.505717, duration: 61.019309\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.325088, val_frame_loss: 2.239066\n",
      "INFO:root:Epoch 3: , nb_train_samples: 1423, train_frame_metric: 0.343625, train_frame_loss: 2.025612, duration: 60.574237\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.391501, val_frame_loss: 1.939204\n",
      "INFO:root:Epoch 4: , nb_train_samples: 1423, train_frame_metric: 0.430120, train_frame_loss: 1.679478, duration: 60.438975\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.441137, val_frame_loss: 1.867528\n",
      "INFO:root:Epoch 5: , nb_train_samples: 1423, train_frame_metric: 0.491723, train_frame_loss: 1.451185, duration: 61.163320\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.510834, val_frame_loss: 1.703157\n",
      "INFO:root:Epoch 6: , nb_train_samples: 1423, train_frame_metric: 0.550546, train_frame_loss: 1.224274, duration: 60.180180\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.522627, val_frame_loss: 1.743494\n",
      "INFO:root:Epoch 7: , nb_train_samples: 1423, train_frame_metric: 0.591495, train_frame_loss: 1.101929, duration: 62.104371\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.496331, val_frame_loss: 1.844608\n",
      "INFO:root:Epoch 8: , nb_train_samples: 1423, train_frame_metric: 0.633822, train_frame_loss: 0.949475, duration: 59.308904\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.516255, val_frame_loss: 1.634175\n",
      "INFO:root:Epoch 9: , nb_train_samples: 1423, train_frame_metric: 0.679030, train_frame_loss: 0.826107, duration: 59.996453\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.545810, val_frame_loss: 1.695186\n",
      "INFO:root:Epoch 10: , nb_train_samples: 1423, train_frame_metric: 0.715961, train_frame_loss: 0.706132, duration: 63.833647\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.586669, val_frame_loss: 1.493656\n",
      "INFO:root:Epoch 11: , nb_train_samples: 1423, train_frame_metric: 0.740236, train_frame_loss: 0.635872, duration: 63.564268\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.585968, val_frame_loss: 1.572260\n",
      "INFO:root:Epoch 12: , nb_train_samples: 1423, train_frame_metric: 0.782323, train_frame_loss: 0.510706, duration: 63.878823\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.582664, val_frame_loss: 1.634705\n",
      "INFO:root:Epoch 13: , nb_train_samples: 1423, train_frame_metric: 0.805098, train_frame_loss: 0.461816, duration: 63.180824\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.604007, val_frame_loss: 1.438769\n",
      "INFO:root:Epoch 14: , nb_train_samples: 1423, train_frame_metric: 0.827925, train_frame_loss: 0.402004, duration: 63.558309\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.609421, val_frame_loss: 1.462118\n",
      "INFO:root:Epoch 15: , nb_train_samples: 1423, train_frame_metric: 0.837262, train_frame_loss: 0.372451, duration: 59.360174\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.629010, val_frame_loss: 1.266039\n",
      "INFO:root:Epoch 16: , nb_train_samples: 1423, train_frame_metric: 0.857746, train_frame_loss: 0.328534, duration: 61.753793\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.637518, val_frame_loss: 1.255563\n",
      "INFO:root:Epoch 17: , nb_train_samples: 1423, train_frame_metric: 0.863770, train_frame_loss: 0.315448, duration: 63.140369\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.633970, val_frame_loss: 1.299203\n",
      "INFO:root:Epoch 18: , nb_train_samples: 1423, train_frame_metric: 0.879408, train_frame_loss: 0.274238, duration: 62.025031\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.631313, val_frame_loss: 1.261005\n",
      "INFO:root:Epoch 19: , nb_train_samples: 1423, train_frame_metric: 0.883284, train_frame_loss: 0.262217, duration: 61.775835\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.626255, val_frame_loss: 1.422362\n",
      "INFO:root:Epoch 20: , nb_train_samples: 1423, train_frame_metric: 0.894971, train_frame_loss: 0.236570, duration: 62.957998\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.627467, val_frame_loss: 1.407359\n",
      "INFO:root:Epoch 21: , nb_train_samples: 1423, train_frame_metric: 0.901388, train_frame_loss: 0.217850, duration: 61.592798\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.640351, val_frame_loss: 1.365421\n",
      "INFO:root:Epoch 22: , nb_train_samples: 1423, train_frame_metric: 0.915150, train_frame_loss: 0.190600, duration: 61.279354\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.654655, val_frame_loss: 1.289390\n",
      "INFO:root:Epoch 23: , nb_train_samples: 1423, train_frame_metric: 0.921962, train_frame_loss: 0.172429, duration: 60.975880\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.646947, val_frame_loss: 1.300395\n",
      "INFO:root:Epoch 24: , nb_train_samples: 1423, train_frame_metric: 0.925071, train_frame_loss: 0.166860, duration: 61.327692\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.652924, val_frame_loss: 1.303838\n",
      "INFO:root:Epoch 25: , nb_train_samples: 1423, train_frame_metric: 0.929814, train_frame_loss: 0.155685, duration: 61.737912\n",
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.649809, val_frame_loss: 1.331862\n"
     ]
    }
   ],
   "source": [
    "val_frame_res, train_frame_res = fit_generator(training_generator=training_batch_generator_obj, validation_generator=testing_batch_generator_obj)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Layer inputs static shape:  (?, 650, 64)\n",
      "Logits static shape:  (?, 650, 48)\n",
      "vars_with_grad:  [<tf.Variable 'encoder_rnn/cell_0/bidirectional_rnn/fw/gru_cell/gates/kernel:0' shape=(320, 512) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_0/bidirectional_rnn/fw/gru_cell/gates/bias:0' shape=(512,) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_0/bidirectional_rnn/fw/gru_cell/candidate/kernel:0' shape=(320, 256) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_0/bidirectional_rnn/fw/gru_cell/candidate/bias:0' shape=(256,) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_0/bidirectional_rnn/bw/gru_cell/gates/kernel:0' shape=(320, 512) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_0/bidirectional_rnn/bw/gru_cell/gates/bias:0' shape=(512,) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_0/bidirectional_rnn/bw/gru_cell/candidate/kernel:0' shape=(320, 256) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_0/bidirectional_rnn/bw/gru_cell/candidate/bias:0' shape=(256,) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_1/bidirectional_rnn/fw/gru_cell/gates/kernel:0' shape=(768, 512) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_1/bidirectional_rnn/fw/gru_cell/gates/bias:0' shape=(512,) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_1/bidirectional_rnn/fw/gru_cell/candidate/kernel:0' shape=(768, 256) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_1/bidirectional_rnn/fw/gru_cell/candidate/bias:0' shape=(256,) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_1/bidirectional_rnn/bw/gru_cell/gates/kernel:0' shape=(768, 512) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_1/bidirectional_rnn/bw/gru_cell/gates/bias:0' shape=(512,) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_1/bidirectional_rnn/bw/gru_cell/candidate/kernel:0' shape=(768, 256) dtype=float32_ref>, <tf.Variable 'encoder_rnn/cell_1/bidirectional_rnn/bw/gru_cell/candidate/bias:0' shape=(256,) dtype=float32_ref>, <tf.Variable 'encoder_time_fully_connected/kernel:0' shape=(512, 48) dtype=float32_ref>, <tf.Variable 'encoder_time_fully_connected/bias:0' shape=(48,) dtype=float32_ref>]\n",
      "INFO:tensorflow:Restoring parameters from ../data/breakfast/results/multilabel_rnn_model.ckpt-2249\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Restoring parameters from ../data/breakfast/results/multilabel_rnn_model.ckpt-2249\n",
      "INFO:root:Number of trainable params: 1698864\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model restored.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:nb_val_samples: 289, val_frame_metric: 0.651600, val_frame_loss: 1.450031\n"
     ]
    }
   ],
   "source": [
    "checkpoint_file = '../data/breakfast/results/multilabel_rnn_model.ckpt-2249'\n",
    "val_frame_res = restore_and_eval(checkpoint_file, training_generator=training_batch_generator_obj, validation_generator=testing_batch_generator_obj)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 7) Visualize results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "val_frame_y_pred = val_frame_res.val_frame_y_pred\n",
    "val_frame_y_true = val_frame_res.val_frame_y_true\n",
    "val_frame_sample_weights = val_frame_res.val_frame_sample_weights\n",
    "\n",
    "nb_val_sequences = val_frame_y_true.shape[0]\n",
    "val_frame_y_true_lst = []\n",
    "val_frame_y_pred_lst = []\n",
    "\n",
    "for seq in range(nb_val_sequences):\n",
    "    val_frame_y_true_lst.append(\n",
    "        val_frame_y_true[seq][nonzero(val_frame_sample_weights[seq])])\n",
    "    val_frame_y_pred_lst.append(\n",
    "        val_frame_y_pred[seq][nonzero(val_frame_sample_weights[seq])])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Visualizing results for testing sequence 15, P07_stereo01_P07_juice (acc: 86.25%)\n",
      "Ground truth labels:\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Choose testing sequence index (0-288)\n",
    "# val_seq = 5  \n",
    "# val_seq = 200\n",
    "# val_seq = 163\n",
    "# val_seq = 269\n",
    "# val_seq = 20\n",
    "# val_seq = 250\n",
    "# val_seq = 111\n",
    "# val_seq = 42\n",
    "# val_seq = 16\n",
    "val_seq = 50\n",
    "val_seq = 18\n",
    "val_seq = 15\n",
    "\n",
    "cmap='gist_ncar'\n",
    "y_true = val_frame_y_true_lst[val_seq]\n",
    "y_pred = val_frame_y_pred_lst[val_seq]\n",
    "video_name = testing_set[val_seq]['video_name']\n",
    "acc = mean(y_true == y_pred) * 100\n",
    "print('Visualizing results for testing sequence {}, {} (acc: {:.02f}%)'.format(val_seq, video_name, acc))\n",
    "plot_label_seq(y_true, nb_classes, actions=action_names,\n",
    "                   cmap=cmap, output_file='../data/breakfast/results/y_true.png', legend=True)\n",
    "plot_label_seq(y_pred, nb_classes, actions=action_names,\n",
    "                   cmap=cmap, output_file='../data/breakfast/results/y_pred.png', legend=True)\n",
    "print('Ground truth labels:')\n",
    "Image(filename='../data/breakfast/results/y_true.png')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predicted labels:\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print('Predicted labels:')\n",
    "Image(filename='../data/breakfast/results/y_pred.png') "
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}