Static Features Classifier

This is a static features classifier for Point-Could clusters using an Attention-RNN model

Installation

$ git clone https://github.com/Mohtasib/Static_Features_Classifier.git
$ cd Static_Features_Classifier
$ conda create -n sfc_env python==3.6.10
$ conda activate sfc_env
$ pip install -e .

Example Code

You can simply run the following command to execute a sample code

$ python examples/train_and_evaluate.py

Usage

1. Import Libraries:

import numpy as np
from sfc.util import create_dataset
from sfc.models.Attention_RNN import Attention_RNN

2. Define the constants:

MAX_SEQ_LEN = ...
NUM_FEATURES = ...

3. Define the directories:

DATA_PATH = ...
My_Model_Logs_DIR = ...
My_Model_Weights = My_Model_Logs_DIR + 'Best_Attention_RNN_ckpt.h5'

4. Create the model:

My_Model = Attention_RNN(Logs_DIR=My_Model_Logs_DIR)

5. Create the dataset and load it into the model:

My_Model.x_train, My_Model.y_train = create_dataset(DATA_PATH + 'train/', MAX_SEQ_LEN, NUM_FEATURES)
My_Model.x_test, My_Model.y_test = create_dataset(DATA_PATH + 'test/', MAX_SEQ_LEN, NUM_FEATURES)

7. Train the model:

My_Model.Fit()

8. Evaluate the model:

My_Model.Evaluate()

9. Predict using the model:

predictions = My_Model.Predict(data)

NOTE: More details will be available soon!!!