Underwater Fish Recognition Model

An underwater fish recognition model developed using YOLOv8. The model was trained on enhanced images of the dataset. The enhancement technique used was RGHS.

Dataset

DeepFish : https://alzayats.github.io/DeepFish/

Dataset enhancement technique RGHS usage

1. List of libraries you need to install to execute the code :

   python = 3.6, cv2, numpy, scipy, matplotlib, scikit-image, natsort, math, datetime

2. Put the input images to corresponding folders :

   Create 'InputImages' and 'OutputImages' folders

   Then put raw images to 'InputImages' folder Figure-Enhanced image after RGHS

3. Run main.py;

The enhanced/restored images will be saved in "OutputImages" folder.

Model Usage

1) Using the deployed model on Roboflow

The model is deployed on Roboflow and can be used directly through following link

https://universe.roboflow.com/underwater-fish/underwater-fish-detection-izi1l/model/6

2) Through Python

Download the model weights file fish_detection_model.pt

Install ultralytics

pip install ultralytics

The fllowing should be run in the python file

from ultralytics import YOLO

# load model
model = YOLO('fish_detection_model.pt')

# predict on image
model.predict('image_file.jpg', save=True, conf=0.5)

Figure- Detection of caranx fish