Future direction(s) in Aquaculture ...

Figure 6. Comparison of recognition results of improved YOLOv5s model before and after enhancement: (a) original image, (b) recognition results of the original image, (c) enhanced image, (d) recognition results of the enhanced image.

 

Crab aquaculture in China spans over 6,000 km², but current manual monitoring methods are inefficient. This study proposes an automated crab detection system using enhanced underwater image processing - including optimized dark channel prior techniques and color correctionpaired with the ShuffleNetV2 model for improved detection speed and accuracy. The method achieves a 90.78% detection rate and significantly outperforms YOLOv5s, offering a practical tool for real-time  - crab monitoring and aquaculture management. ( https://www.mdpi.com/2410-3888/9/2/60 )