Fish detection and segmentation based on YOLOv3 that use GrabCut to do semantic segmentation to fish market images. Trained by FISH9004
You can see the DEMO here
To downloand the image and run the contaider in detach mode, run the code below.
docker container run -p 8501:8501 --rm -d pablogod/fishv4To shutdown the docker type this:
docker ps -aq # Check which id was assigned for the fishv4 instance
docker stop <weird id of fishv4> # Type the idLocally:
git clone https://github.com/DZPeru/fishv4
cd fishv4
pip3 install -r requirements.txt
gdown --output ./fishv4/fish.weights --id 1QxJHz8Lh4CRE-4ksitJkLdpOpzGFkg-lConda:
conda create -n fishv4 python=3.6 pip
conda activate fishv4
pip install -r requirements.txtDownload the weights of the neural network to your local repository. Or do it manually, downloading from Google Drive.
gdown --output ./fishv4/fish.weights --id 1QxJHz8Lh4CRE-4ksitJkLdpOpzGFkg-lpython main.py --image ./path_to/my_image.jpg --yolo yolov4When finishing, you should find images (.jpg) in the project root directory.
streamlit run app.pyYou can upload fish market image to run the program. The results are shown in the browser (make sure to scroll down).