SmartFilter

How to run SmartFilter

1. Clone the repo and make sure you have installed the require dependancies from the requirements.txt file. For GPU users make sure you install the GPU version of the following libraries: tensorflow,torch and detectron2
2. Run following commands to download and prepare the video dataset for supervised traning

cd  ./SmartFilter
python -m   src.download_dataset --data_path ./data

./data is the path to where the downloaded data will be stored

Supervised Training

 python -m  src.supervised_training --data_path ./data

Online Adaptation

1. Start the server:

    python -m src.server --adapt True 

2. Stream from the camera to server

    python -m src.camera --ip SERVER_IP

3. Stream a pre-recoded video to the server

    python -m src.camera --ip SERVER_IP --video ./video.mp4

Live Filtering

1. Server:

    python -m src.server --stream_to_client True 

2. Camera:

    python -m src.camera --filter True 

3. Client:

    python -m src.client --ip SERVER_IP