edwin-19/Fish-Recognizer

YOLOV3 as detector and EffcientNET(planto) for classification

Fish Recognizer

Dependencies

• Keras
• Tensorflow (1.9)
• Opencv
• PIL

How to use

Clone this respository

git clone https://github.com/edwin-19/Fish-Recognizer.git

Donwload weights from here

Then change path for weights in yolo.py for files

Then run the following command

python yolo_video.py --input input/out.mp4 --output output/test.avi

Sample Output

TODO

• - Increase accuracy on detector
• - Add SORT Algorithm
• - Add recognizer for fish

Acknowledgement

This repo was heavily insipred by the following repos do check out their work if your interested:

• https://github.com/qqwweee/keras-yolo3
• https://github.com/Qidian213/deep_sort_yolov3

I borrowed the weights trained by the following repo to label some of the data:

• https://github.com/kwea123/fish_detection