Dev

.github/workflows/release.yml

@@ -20,29 +20,26 @@ jobs:
       - name: Install dependencies
         run: |
           python -m pip install --upgrade pip
-          pip install -e .[testing]
-          pip install matplotlib numpy opencv-python pdbpp tensorboard tensorflow-cpu scipy Pillow gdown isort black flake8
-          pip install mypy types-requests pytype pytest pytest-cov pre-commit pytest-mock pytest-flask
+          pip install -e .[tfcpu,api,dev,testing,linting]
 
       - name: Test with pytest
         run: |
           pip install python-coveralls
-          python -m pytest --cov=./dfp --cov-report term-missing
-          pip install coveragepy-lcov
-          coveragepy-lcov --data_file_path .coverage --output_file_path lcov.info
+          python -m pytest --cov=./dfp --cov-report lcov:lcov.info
 
       - name: Build
         run: |
-          pip install setuptools sdist wheel twine
-          pip install -r requirements.txt
-          python setup.py sdist bdist_wheel
+          python -m pip install --upgrade pip
+          pip install -e .[tfcpu,api]
+
       - uses: "marvinpinto/action-automatic-releases@latest"
         with:
           repo_token: "${{ secrets.GITHUB_TOKEN }}"
           prerelease: false
           files: |
             ./dist/*tar.gz
             ./dist/*.whl
+
       - name: Publish distribution 📦 to PyPI
         uses: pypa/gh-action-pypi-publish@master
         with:

.gitignore

@@ -146,3 +146,4 @@ model/
 cov.info
 lcov.info
 *.h5
+out.jpg

Dockerfile

@@ -18,7 +18,8 @@ RUN pip install -e .[tfgpu,api]
 # RUN gdown https://drive.google.com/uc?id=1czUSFvk6Z49H-zRikTc67g2HUUz4imON
 # RUN unzip log.zip
 # RUN rm log.zip
-ADD log log
+COPY docs/app.toml /docs/app.toml
+ADD log/store log/store
 
 COPY resources /usr/local/resources
 RUN mv /usr/local/resources .

README.md

@@ -1,5 +1,4 @@
-# TF2DeepFloorplan [![License: GPL v3](https://img.shields.io/badge/License-GPLv3-blue.svg)](https://www.gnu.org/licenses/gpl-3.0) [<img src="https://colab.research.google.com/assets/colab-badge.svg" >](https://colab.research.google.com/github/zcemycl/TF2DeepFloorplan/blob/master/deepfloorplan.ipynb) ![example workflow](https://github.com/zcemycl/TF2DeepFloorplan/actions/workflows/main.yml/badge.svg) [![Coverage Status](https://coveralls.io/repos/github/zcemycl/TF2DeepFloorplan/badge.svg?branch=main)](https://coveralls.io/github/zcemycl/TF2DeepFloorplan?branch=main)
-<!-- [![Hits](https://hits.seeyoufarm.com/api/count/incr/badge.svg?url=https%3A%2F%2Fgithub.com%2Fzcemycl%2FTF2DeepFloorplan&count_bg=%2379C83D&title_bg=%23555555&icon=&icon_color=%23E7E7E7&title=hits&edge_flat=false)](https://hits.seeyoufarm.com) -->
+# TF2DeepFloorplan [![License: GPL v3](https://img.shields.io/badge/License-GPLv3-blue.svg)](https://www.gnu.org/licenses/gpl-3.0) [<img src="https://colab.research.google.com/assets/colab-badge.svg" >](https://colab.research.google.com/github/zcemycl/TF2DeepFloorplan/blob/master/deepfloorplan.ipynb) ![example workflow](https://github.com/zcemycl/TF2DeepFloorplan/actions/workflows/main.yml/badge.svg) [![Coverage Status](https://coveralls.io/repos/github/zcemycl/TF2DeepFloorplan/badge.svg?branch=main)](https://coveralls.io/github/zcemycl/TF2DeepFloorplan?branch=main)[![Hits](https://hits.seeyoufarm.com/api/count/incr/badge.svg?url=https%3A%2F%2Fgithub.com%2Fzcemycl%2FTF2DeepFloorplan&count_bg=%2379C83D&title_bg=%23555555&icon=&icon_color=%23E7E7E7&title=hits&edge_flat=false)](https://hits.seeyoufarm.com)
 This repo contains a basic procedure to train and deploy the DNN model suggested by the paper ['Deep Floor Plan Recognition using a Multi-task Network with Room-boundary-Guided Attention'](https://arxiv.org/abs/1908.11025). It rewrites the original codes from [zlzeng/DeepFloorplan](https://github.com/zlzeng/DeepFloorplan) into newer versions of Tensorflow and Python.
 <br>
 Network Architectures from the paper, <br>
@@ -36,8 +35,10 @@ pip install -e .[tfgpu,api,dev,testing,linting]
 ```
 python -m dfp.train [--batchsize 2][--lr 1e-4][--epochs 1000]
 [--logdir 'log/store'][--modeldir 'model/store']
-[--saveTensorInterval 10][--saveModelInterval 20]
-[--tfmodel 'subclass'/'func']
+[--save-tensor-interval 10][--save-model-interval 20]
+[--tfmodel 'subclass'/'func'][--feature-channels 256 128 64 32]
+[--backbone 'vgg16'/'mobilenetv1'/'mobilenetv2'/'resnet50']
+[--feature-names block1_pool block2_pool block3_pool block4_pool block5_pool]
 ```
 - for example,
 ```
@@ -52,7 +53,11 @@ tensorboard --logdir=log/store
 ```
 python -m dfp.convert2tflite [--modeldir model/store]
 [--tflitedir model/store/model.tflite]
-[--quantize]
+[--loadmethod 'log'/'none'/'pb']
+[--quantize][--tfmodel 'subclass'/'func']
+[--feature-channels 256 128 64 32]
+[--backbone 'vgg16'/'mobilenetv1'/'mobilenetv2'/'resnet50']
+[--feature-names block1_pool block2_pool block3_pool block4_pool block5_pool]
 ```
 6. Download and unzip model from google drive,
 ```
@@ -70,6 +75,9 @@ python -m dfp.deploy [--image 'path/to/image']
 [--loadmethod 'log'/'pb'/'tflite']
 [--weight 'log/store/G'/'model/store'/'model/store/model.tflite']
 [--tfmodel 'subclass'/'func']
+[--feature-channels 256 128 64 32]
+[--backbone 'vgg16'/'mobilenetv1'/'mobilenetv2'/'resnet50']
+[--feature-names block1_pool block2_pool block3_pool block4_pool block5_pool]
 ```
 - for example,
 ```
@@ -83,16 +91,20 @@ python -m dfp.deploy --image floorplan.jpg --weight log/store/G
 docker build -t tf_docker -f Dockerfile .
 docker run -d -p 1111:1111 tf_docker:latest
 docker run --gpus all -d -p 1111:1111 tf_docker:latest
+
+# special for hot reloading flask
+docker run -v ${PWD}/src/dfp/app.py:/src/dfp/app.py -v ${PWD}/src/dfp/deploy.py:/src/dfp/deploy.py -d -p 1111:1111 tf_docker:latest
+docker logs `docker ps | grep "tf_docker:latest"  | awk '{ print $1 }'` --follow
 ```
 2. Call the api for output.
 ```
 curl -H "Content-Type: application/json" --request POST  \
-  -d '{"uri":"https://cdn.cnn.com/cnnnext/dam/assets/200212132008-04-london-rental-market-intl-exlarge-169.jpg","colorize":1,"postprocess":0, "output":"/tmp"}' \
-  http://0.0.0.0:1111/process --output /tmp/tmp.jpg
+  -d '{"uri":"https://cdn.cnn.com/cnnnext/dam/assets/200212132008-04-london-rental-market-intl-exlarge-169.jpg","colorize":1,"postprocess":0}' \
+  http://0.0.0.0:1111/uri --output /tmp/tmp.jpg
 
 
-curl --request POST -F "file=@resources/30939153.jpg;type=image/jpeg" \
-  -F "postprocess=0" -F "colorize=0" -F "output=/tmp" http://0.0.0.0:1111/process --output out.jpg
+curl --request POST -F "file=@resources/30939153.jpg" \
+  -F "postprocess=0" -F "colorize=0" http://0.0.0.0:1111/upload --output out.jpg
 ```
 3. If you run `app.py` without docker, the second curl for file upload will not work.
 
@@ -148,27 +160,27 @@ git branch -d xx-features
 ## Optimization
 - Backbone Comparison in Size
 
-|Backbone|log|pb|tflite|
-|---|---|---|---|
-|VGG16|130.5Mb|119Mb|45.3Mb|
-|MobileNetV1|102.1Mb|86.7Mb|50.2Mb|
-|MobileNetV2|129.3Mb|94.4Mb|57.9Mb|
-|ResNet50|214Mb|216Mb|107.2Mb|
+|Backbone|log|pb|tflite|toml|
+|---|---|---|---|---|
+|VGG16|130.5Mb|119Mb|45.3Mb|[link](docs/experiments/vgg16/exp1)|
+|MobileNetV1|102.1Mb|86.7Mb|50.2Mb|[link](docs/experiments/mobilenetv1/exp1)|
+|MobileNetV2|129.3Mb|94.4Mb|57.9Mb|[link](docs/experiments/mobilenetv2/exp1)|
+|ResNet50|214Mb|216Mb|107.2Mb|[link](docs/experiments/resnet50/exp1)|
 
 - Feature Selection Comparison in Size
 
-|Backbone|Feature Names|log|pb|tflite|
-|---|---|---|---|---|
-|MobileNetV1|"conv_pw_1_relu", <br>"conv_pw_3_relu", <br>"conv_pw_5_relu", <br>"conv_pw_7_relu", <br>"conv_pw_13_relu"|102.1Mb|86.7Mb|50.2Mb|
-|MobileNetV1|"conv_pw_1_relu", <br>"conv_pw_3_relu", <br>"conv_pw_5_relu", <br>"conv_pw_7_relu", <br>"conv_pw_12_relu"|84.5Mb|82.3Mb|49.2Mb|
+|Backbone|Feature Names|log|pb|tflite|toml|
+|---|---|---|---|---|---|
+|MobileNetV1|"conv_pw_1_relu", <br>"conv_pw_3_relu", <br>"conv_pw_5_relu", <br>"conv_pw_7_relu", <br>"conv_pw_13_relu"|102.1Mb|86.7Mb|50.2Mb|[link](docs/experiments/mobilenetv1/exp1)|
+|MobileNetV1|"conv_pw_1_relu", <br>"conv_pw_3_relu", <br>"conv_pw_5_relu", <br>"conv_pw_7_relu", <br>"conv_pw_12_relu"|84.5Mb|82.3Mb|49.2Mb|[link](docs/experiments/mobilenetv1/exp2)|
 
 - Feature Channels Comparison in Size
 
-|Backbone|Channels|log|pb|tflite|
-|---|---|---|---|---|
-|VGG16|[256,128,64,32]|130.5Mb|119Mb|45.3Mb|
-|VGG16|[128,64,32,16]|82.4Mb|81.6Mb|27.3Mb|
-|VGG16|[32,32,32,32]|73.2Mb|67.5Mb|18.1Mb|
+|Backbone|Channels|log|pb|tflite|toml|
+|---|---|---|---|---|---|
+|VGG16|[256,128,64,32]|130.5Mb|119Mb|45.3Mb|[link](docs/experiments/vgg16/exp1)|
+|VGG16|[128,64,32,16]|82.4Mb|81.6Mb|27.3Mb||
+|VGG16|[32,32,32,32]|73.2Mb|67.5Mb|18.1Mb|[link](docs/experiments/vgg16/exp2)|
 
 - tfmot
   - Pruning (not working)

deepfloorplan.ipynb

@@ -74,6 +74,7 @@
     "from dfp.utils.rgb_ind_convertor import *\n",
     "from dfp.utils.util import *\n",
     "from dfp.utils.legend import *\n",
+    "from dfp.utils.settings import *\n",
     "from dfp.deploy import *\n",
     "print(tf.test.is_gpu_available())\n",
     "print(tf.config.list_physical_devices('GPU'))"
@@ -87,7 +88,11 @@
    },
    "outputs": [],
    "source": [
-    "inp = mpimg.imread('./TF2DeepFloorplan/resources/30939153.jpg')"
+    "img_path = './TF2DeepFloorplan/resources/30939153.jpg'\n",
+    "inp = mpimg.imread(img_path)\n",
+    "args = parse_args(\"--tomlfile ./TF2DeepFloorplan/docs/notebook.toml\".split())\n",
+    "args = overwrite_args_with_toml(args)\n",
+    "args.image = img_path"
    ]
   },
   {
@@ -102,10 +107,6 @@
    },
    "outputs": [],
    "source": [
-    "args = Namespace(image='./TF2DeepFloorplan/resources/30939153.jpg',\n",
-    "        weight='./log/store/G',loadmethod='log',\n",
-    "        postprocess=True,colorize=True,\n",
-    "        save=None,tfmodel=\"subclass\")\n",
     "result = main(args)"
    ]
   },

docs/app.toml

@@ -0,0 +1,10 @@
+tfmodel = 'subclass'
+image = ''
+postprocess = 1
+colorize = 1
+save = ''
+weight = 'log/store/G'
+loadmethod = 'log'
+feature_channels = [256, 128, 64, 32]
+backbone = 'vgg16'
+feature_names = ["block1_pool", "block2_pool", "block3_pool", "block4_pool", "block5_pool",]

docs/experiments/mobilenetv1/exp1/compress.toml

@@ -0,0 +1,9 @@
+tfmodel = 'func'
+modeldir = 'model/store_mobilenetv1_exp1'
+tflitedir = 'model/store_mobilenetv1_exp1_model.tflite'
+quantize = 1
+compress_mode = 'quantization'
+loadmethod = 'pb'
+feature_channels = [256, 128, 64, 32]
+backbone = 'mobilenetv1'
+feature_names = ["conv_pw_1_relu", "conv_pw_3_relu", "conv_pw_5_relu", "conv_pw_7_relu", "conv_pw_13_relu",]

docs/experiments/mobilenetv1/exp1/deploy.toml

@@ -0,0 +1,14 @@
+tfmodel = 'func'
+image = 'resources/30939153.jpg'
+postprocess = 1
+colorize = 1
+save = ''
+weight = 'log/store_mobilenetv1_exp1/G'
+loadmethod = 'log'
+# weight = 'model/store_mobilenetv1_exp1'
+# loadmethod = 'pb'
+# weight = 'model/store_mobilenetv1_exp1_model.tflite'
+# loadmethod = 'tflite'
+feature_channels = [256, 128, 64, 32]
+backbone = 'mobilenetv1'
+feature_names = ["conv_pw_1_relu", "conv_pw_3_relu", "conv_pw_5_relu", "conv_pw_7_relu", "conv_pw_13_relu",]

docs/experiments/mobilenetv1/exp1/train.toml

@@ -0,0 +1,13 @@
+tfmodel = 'func'
+batchsize = 8
+lr = 1e-4
+wd = 1e-5
+epochs = 100
+logdir = 'log/store_mobilenetv1_exp1'
+modeldir = 'model/store_mobilenetv1_exp1'
+weight = ''
+save_tensor_interval = 10
+save_model_interval = 20
+feature_channels = [256, 128, 64, 32]
+backbone = 'mobilenetv1'
+feature_names = ["conv_pw_1_relu", "conv_pw_3_relu", "conv_pw_5_relu", "conv_pw_7_relu", "conv_pw_13_relu",]

docs/experiments/mobilenetv1/exp2/compress.toml

@@ -0,0 +1,9 @@
+tfmodel = 'func'
+modeldir = 'model/store_mobilenetv1_exp2'
+tflitedir = 'model/store_mobilenetv1_exp2_model.tflite'
+quantize = 1
+compress_mode = 'quantization'
+loadmethod = 'pb'
+feature_channels = [256, 128, 64, 32]
+backbone = 'mobilenetv1'
+feature_names = ["conv_pw_1_relu", "conv_pw_3_relu", "conv_pw_5_relu", "conv_pw_7_relu", "conv_pw_12_relu",]

docs/experiments/mobilenetv1/exp2/deploy.toml

@@ -0,0 +1,14 @@
+tfmodel = 'func'
+image = 'resources/30939153.jpg'
+postprocess = 1
+colorize = 1
+save = ''
+weight = 'log/store_mobilenetv1_exp2/G'
+loadmethod = 'log'
+# weight = 'model/store_mobilenetv1_exp2'
+# loadmethod = 'pb'
+# weight = 'model/store_mobilenetv1_exp2_model.tflite'
+# loadmethod = 'tflite'
+feature_channels = [256, 128, 64, 32]
+backbone = 'mobilenetv1'
+feature_names = ["conv_pw_1_relu", "conv_pw_3_relu", "conv_pw_5_relu", "conv_pw_7_relu", "conv_pw_12_relu",]

docs/experiments/mobilenetv1/exp2/train.toml

@@ -0,0 +1,13 @@
+tfmodel = 'func'
+batchsize = 8
+lr = 1e-4
+wd = 1e-5
+epochs = 100
+logdir = 'log/store_mobilenetv1_exp2'
+modeldir = 'model/store_mobilenetv1_exp2'
+weight = ''
+save_tensor_interval = 10
+save_model_interval = 20
+feature_channels = [256, 128, 64, 32]
+backbone = 'mobilenetv1'
+feature_names = ["conv_pw_1_relu", "conv_pw_3_relu", "conv_pw_5_relu", "conv_pw_7_relu", "conv_pw_12_relu",]

docs/experiments/mobilenetv2/exp1/train.toml

@@ -0,0 +1,13 @@
+tfmodel = 'func'
+batchsize = 8
+lr = 1e-4
+wd = 1e-5
+epochs = 100
+logdir = 'log/store_mobilenetv2_exp1'
+modeldir = 'model/store_mobilenetv2_exp1'
+weight = ''
+save_tensor_interval = 10
+save_model_interval = 20
+feature_channels = [256, 128, 64, 32]
+backbone = 'mobilenetv2'
+feature_names = ["block_1_expand_relu", "block_3_expand_relu", "block_5_expand_relu", "block_13_expand_relu", "out_relu",]

docs/experiments/resnet50/exp1/train.toml

@@ -0,0 +1,13 @@
+tfmodel = 'func'
+batchsize = 8
+lr = 1e-4
+wd = 1e-5
+epochs = 100
+logdir = 'log/store_resnet50_exp1'
+modeldir = 'model/store_resnet50_exp1'
+weight = ''
+save_tensor_interval = 10
+save_model_interval = 20
+feature_channels = [256, 128, 64, 32]
+backbone = 'resnet50'
+feature_names = ["conv1_relu", "conv2_block3_out", "conv3_block4_out", "conv4_block6_out", "conv5_block3_out",]

docs/experiments/vgg16/exp1/compress.toml

@@ -0,0 +1,9 @@
+tfmodel = 'subclass'
+modeldir = 'model/store_vgg16_exp1'
+tflitedir = 'model/store_vgg16_exp1_model.tflite'
+quantize = 1
+compress_mode = 'quantization'
+loadmethod = 'pb'
+feature_channels = [256, 128, 64, 32]
+backbone = 'vgg16'
+feature_names = ["block1_pool", "block2_pool", "block3_pool", "block4_pool", "block5_pool",]

docs/experiments/vgg16/exp1/compress_log.toml

@@ -0,0 +1,9 @@
+tfmodel = 'subclass'
+modeldir = 'log/store_vgg16_exp1/G'
+tflitedir = 'model/store_vgg16_exp1_model_log.tflite'
+quantize = 1
+compress_mode = 'quantization'
+loadmethod = 'log'
+feature_channels = [256, 128, 64, 32]
+backbone = 'vgg16'
+feature_names = ["block1_pool", "block2_pool", "block3_pool", "block4_pool", "block5_pool",]

docs/experiments/vgg16/exp1/deploy.toml

@@ -0,0 +1,14 @@
+tfmodel = 'subclass'
+image = 'resources/30939153.jpg'
+postprocess = 1
+colorize = 1
+save = ''
+# weight = 'log/store_vgg16_exp1/G'
+# loadmethod = 'log'
+# weight = 'model/store_vgg16_exp1'
+# loadmethod = 'pb'
+weight = 'model/store_vgg16_exp1_model.tflite'
+loadmethod = 'tflite'
+feature_channels = [256, 128, 64, 32]
+backbone = 'vgg16'
+feature_names = ["block1_pool", "block2_pool", "block3_pool", "block4_pool", "block5_pool",]

docs/experiments/vgg16/exp1/train.toml

@@ -0,0 +1,13 @@
+tfmodel = 'subclass'
+batchsize = 8
+lr = 1e-4
+wd = 1e-5
+epochs = 100
+logdir = 'log/store_vgg16_exp1'
+modeldir = 'model/store_vgg16_exp1'
+weight = ''
+save_tensor_interval = 10
+save_model_interval = 20
+feature_channels = [256, 128, 64, 32]
+backbone = 'vgg16'
+feature_names = ["block1_pool", "block2_pool", "block3_pool", "block4_pool", "block5_pool",]

docs/experiments/vgg16/exp2/compress.toml

@@ -0,0 +1,9 @@
+tfmodel = 'subclass'
+modeldir = 'model/store_vgg16_exp2'
+tflitedir = 'model/store_vgg16_exp2_model.tflite'
+quantize = 1
+compress_mode = 'quantization'
+loadmethod = 'pb'
+feature_channels = [32, 32, 32, 32]
+backbone = 'vgg16'
+feature_names = ["block1_pool", "block2_pool", "block3_pool", "block4_pool", "block5_pool",]

docs/experiments/vgg16/exp2/deploy.toml

@@ -0,0 +1,14 @@
+tfmodel = 'subclass'
+image = 'resources/30939153.jpg'
+postprocess = 1
+colorize = 1
+save = ''
+weight = 'log/store_vgg16_exp2/G'
+loadmethod = 'log'
+# weight = 'model/store_vgg16_exp2'
+# loadmethod = 'pb'
+# weight = 'model/store_vgg16_exp2_model.tflite'
+# loadmethod = 'tflite'
+feature_channels = [32, 32, 32, 32]
+backbone = 'vgg16'
+feature_names = ["block1_pool", "block2_pool", "block3_pool", "block4_pool", "block5_pool",]