Fix readme files and SR demo (openvinotoolkit#113)

* Some fixes in readme files

* Add checkpoint ot ignore

* Fix resize

* File tree

Author: AlexanderDokuchaev

.gitignore

@@ -11,3 +11,8 @@ data/*
 external/*
 
 *egg-info
+
+*.pth
+*.onnx
+*.xml
+*.bin

README.md

@@ -32,6 +32,10 @@ inference.
   * [Face recognition](pytorch_toolkit/face_recognition)
   * [Human pose estimation](pytorch_toolkit/human_pose_estimation)
   * [Instance segmentation](pytorch_toolkit/instance_segmentation)
+  * [Object Detection](pytorch_toolkit/object_detection)
+    - [Face Detection](pytorch_toolkit/object_detection/face_detection.md)
+    - [Person Vehicle Bike Detector](pytorch_toolkit/object_detection/person_vehicle_bike_detection.md)
+  * [Segmentation of thoracic organs](pytorch_toolkit/segthor)
   * [Super resolution](pytorch_toolkit/super_resolution)
 
 * [TensorFlow](tensorflow_toolkit)

pytorch_toolkit/README.md

@@ -44,7 +44,12 @@ Do not forget to update several environment variables are required to compile an
 After installation, you are ready to train your own models, evaluate them, use
 them for predictions.
 
+* [Action recognition](action_recognition)
 * [Face recognition](face_recognition)
 * [Human pose estimation](human_pose_estimation)
 * [Instance segmentation](instance_segmentation)
+* [Object Detection](object_detection)
+  - [Face Detection](object_detection/face_detection.md)
+  - [Person Vehicle Bike Detector](object_detection/person_vehicle_bike_detection.md)
+* [Segmentation of thoracic organs](segthor)
 * [Super resolution](super_resolution)

pytorch_toolkit/object_detection/face_detection.md

@@ -5,32 +5,27 @@ This repository contains training scripts for lightweight SSD-based face detecto
 
 ## Prerequisites
 
-1. Download mmdetection submodule:
-  `git submodule update --init --recommend-shallow ../../external/mmdetection`
-2. Download the [WIDER Face](http://shuoyang1213.me/WIDERFACE/) and unpack it to `data` folder.
-3. Annotation in the VOC format can be found in this
+1. Download the [WIDER Face](http://shuoyang1213.me/WIDERFACE/) and unpack it to `data` folder.
+2. Annotation in the VOC format can be found in this
 [repo](https://github.com/sovrasov/wider-face-pascal-voc-annotations.git). Move the annotation files from
 `WIDER_train_annotations` and `WIDER_val_annotations` folders to the `Annotation` folders inside the corresponding
 directories `WIDER_train` and `WIDER_val`. Also annotation lists `val.txt` and `train.txt` should be copied to
 `data/WIDERFace` from `WIDER_train_annotations` and `WIDER_val_annotations`.
 The directory should be like this:
 
 ```
-object_detection
-├── tools
-├── data
-│   ├── WIDERFace
-│   │   ├── WIDER_train
-│   |   │   ├──0--Parade
-│   |   │   ├── ...
-│   |   │   ├── Annotations
-│   │   ├── WIDER_val
-│   |   │   ├──0--Parade
-│   |   │   ├── ...
-│   |   │   ├── Annotations
-│   │   ├── val.txt
-│   │   ├── train.txt
-
+data
+└── WIDERFace
+    ├── WIDER_train
+    │   ├──0--Parade
+    │   ├── ...
+    │   └── Annotations
+    ├── WIDER_val
+    │   ├──0--Parade
+    │   ├── ...
+    │   └── Annotations
+    ├── val.txt
+    └── train.txt
 ```
 
 ## Training
@@ -40,7 +35,7 @@ object_detection
 2. To train the detector on a single GPU run in terminal:
   ```bash
   python3 ../../external/mmdetection/tools/train.py \
-    ../../external/mmdetection/configs/wider_face/mobilenetv2_tiny_ssd300_wider_face.py
+  configs/mobilenetv2_tiny_ssd300_wider_face.py
   ```
 
 
@@ -58,7 +53,7 @@ object_detection
   ```bash
   python3 ../../external/mmdetection/tools/voc_eval.py    \
     result.pkl    \
-    ../../external/mmdetection/configs/wider_face/mobilenetv2_tiny_ssd300_wider_face.py
+    configs/mobilenetv2_tiny_ssd300_wider_face.py
   ```
   One should observe 0.305 AP on validation set. For more detailed results and comparison with vanilla SSD300 see `../../external/mmdetection/configs/wider_face/README.md`.
 
@@ -67,7 +62,7 @@ object_detection
 1. Convert PyTorch model to ONNX format: run script in terminal
   ```bash
   python3 tools/onnx_export.py \
-        ../../external/mmdetection/configs/wider_face/mobilenetv2_tiny_ssd300_wider_face.py
+        configs/mobilenetv2_tiny_ssd300_wider_face.py
         <CHECKPOINT> \
         face_detector.onnx
   ```
@@ -89,7 +84,7 @@ object_detection
 To run the demo connect a webcam end execute command:
 ```bash
 python3 tools/detection_live_demo.py  \
-  ../../external/mmdetection/configs/wider_face/mobilenetv2_tiny_ssd300_wider_face.py \
+  configs/mobilenetv2_tiny_ssd300_wider_face.py \
   <CHECKPOINT> \
   --cam_id 0
 ```
@@ -109,5 +104,5 @@ python3 tools/count_flops.py configs/mobilenetv2_tiny_ssd300_wider_face.py
  described in this instruction.
 * Fine-tuning steps are the same as step 2 for training, but some adjustments in config are needed:
   - specify initial checkpoint containing a valid detector in `load_from` field of config
-    `configsmobilenetv2_tiny_ssd300_wider_face.py`
+    `configs/mobilenetv2_tiny_ssd300_wider_face.py`
   - edit `data` section of config to pass a custom dataset.

pytorch_toolkit/segthor/README.md

@@ -104,8 +104,8 @@ pip3 install -e .
 
 ```
 python tools/prepare_training_dataset.py \
-  --input_path /home/localadmin/segthor/data/train \
-  --output_path /home/localadmin/segthor/data/experimental \
+  --input_path data/train \
+  --output_path data/experimental \
   --new_scale 1 1 2.5
 ```
 

pytorch_toolkit/super_resolution/tools/infer_ie.py

@@ -69,7 +69,7 @@ def main():
     if image.shape[0] != ih or image.shape[1] != iw:
         image = image[0:ih, 0:iw]
 
-    cubic = cv2.resize(image, tuple(inputs[1][1][2:]), interpolation=cv2.INTER_CUBIC)
+    cubic = cv2.resize(image, (inputs[1][1][3], inputs[1][1][2]), interpolation=cv2.INTER_CUBIC)
 
     blob1 = image_to_blob(image, (inputs[0][1]))
     blob2 = image_to_blob(cubic, (inputs[1][1]))