mufaddal12 / fingertip_detection

Setting up the project on your system:

• Install the required packages using pip install -r requirements.txt
• Clone this git repository to your system
• Download trained weights from here into the weights folder
• In file detector.data (in requirements folder) replace the path to the objects.name to that according to your system in names field

• cd into the darknet folder, then
  • If you want to give gpu and cudnn access to the model then run the following two commands on terminal: sed -i 's/GPU=0/GPU=1/' Makefile sed -i 's/CUDNN=0/CUDNN=1/' Makefile
  • Give opencv access by sed -i 's/OPENCV=0/OPENCV=1/' Makefile

For running the model then go to the previous directory cd ..

Run the following commands as per their use:

• For running model on an image run python image_detection.py --imagepath
  • where imagepath is path to image you want to run the detection on
  • the resultant image is saved as "filename_output.jpg" in the detections folder
  • An example image "image.jpg" has been provided in example folder
• For running model on a video, run python video_detection.py --videopath
  • where videopath is path to vidoe you want to run the detection on and
  • the resultant video is saved as "filename_output.avi" in the detections folder
  • An example video "video.webm" has been provided in example folder
• For running live detection from webcam, run python webcam_detection.py //would suggest using gpu for this one