Fishv4 - Fish segmentation

Fish detection and segmentation based on YOLOv3 that use GrabCut to do semantic segmentation to fish market images. Trained by FISH9004

How to run

Web version

You can see the DEMO here

Docker version

To downloand the image and run the contaider in detach mode, run the code below.

docker container run -p 8501:8501 --rm -d pablogod/fishv4

To 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 id

Local computer

Locally:

git clone https://github.com/DZPeru/fishv4
cd fishv4
pip3 install -r requirements.txt
gdown --output ./fishv4/fish.weights --id 1QxJHz8Lh4CRE-4ksitJkLdpOpzGFkg-l

Conda version

Conda:

conda create -n fishv4 python=3.6 pip 
conda activate fishv4
pip install -r requirements.txt

Download 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-l

1. Command Line Approach

python main.py --image ./path_to/my_image.jpg --yolo yolov4

When finishing, you should find images (.jpg) in the project root directory.

2. Streamlit Approach

streamlit run app.py

You can upload fish market image to run the program. The results are shown in the browser (make sure to scroll down).

17853313621/fishv4