FATES-ATI-EnhancedYOLOv8

YOLOv8-TF: Transformer-Enhanced YOLOv8 for Underwater Fish Species Recognition with Class Imbalance Handling

For installing dependencies, pip install ultralytics

For demo training purpose, small dataset consisting around 1000 fish images and labels is provided inside directory : datasets/Small_Train_GFISHERD24

To train in this small dataset for demo purpose following script can be used: python trainYOLOv8_demo.py

Different paths can be given, inside ...dataset_name_.yaml to work with full datasets SEAMAPD21 or GFISHERSD24.

If you find this work useful, please cite our work as:

1. C. Shah, M. M. Nabi, S. Y. Alaba, I. A. Ebu, J. Prior, M. D. Campbell, R. Caillouet, M. D. Grossi, T. Rowell, F. Wallace, J. E. Ball, and R. Moorhead, “YOLOv8-TF: Transformer-Enhanced YOLOv8 for Underwater Fish Species Recognition with Class Imbalance Handling,” MDPI Sensors.
2. Ultralytics, YOLOv8 - State-of-the-art YOLO models, Jan. 2023. [Online]. Available: https://github.com/ultralytics/ultralytics.