rahuja123/Retrieval_Demo

Code for demo GUI person-reid

NTUReID Indoor Demo GUI

We don't support backend functionality right now. Only the frontend GUI can be run. The backend integration will be done soon.

Requirements

• dash by plotly
• flask
• python 3
• pytorch 0.4+
• fire
• opencv-python

Install all dependences libraries

pip3 install -r requirements.txt

To-DO

• need to specify the size of image upload.
• Add the background functionality.

Support Model And Weight:

Model	Weight (Train on Market)
ResNet50	ResNet50_Market
ResNet101	ResNet101_Market
SE_ResNet50	SE_ResNet50_Market
SE_ResNet101	SE_ResNet101_Market

default network ResNet50 default weight ResNet50_Market

Running:

### Follow the code below for GUI. 
python app_exp.py

-- TO-DO (Will be added soon)

### python run.py [target_image] [video_source]
python run.py  ./media/images/jiashu.jpg ./media/videos/S21-B4-L-13.mp4

### python run.py [target_image] [video_source] [model] [weight]
python run.py  ./media/images/jiashu.jpg ./media/videos/S21-B4-L-13.mp4 ResNet101 ResNet101_Market

### python run.py [target_image] [video_source] [threshold] [skip_frame]
python run.py  ./media/images/jiashu.jpg ./media/videos/S21-B4-L-13.mp4 400 5

### python run.py [target_image] [video_source] [threshold] [skip_frame] [yolo_device] [reid_device]
python run.py  ./media/images/jiashu.jpg ./media/videos/S21-B4-L-13.mp4 --yolo_device='cuda:0' --reid_device='cuda:1'