Comparative analysis of YOLOv5 and YOLOv7 for Underwater Object Detection Cases
This research study intends to evaluate and compare the performance of YOLOv5 and YOLOv7 utilizing the trashcan and brackish datasets in order to design an optimal underwater object recognition system for the ROV BlueRov2. Experimental settings were created to assess how well these algorithms performed with various equipment arrangements, particularly in murky environments. Using YOLOv5 and YOLOv7, a lightweight object identification approach was presented to overcome the difficulties in underwater object detection, such as low visibility, color bias, and small targets. This approach attained a high mean average precision (mAP). Moreover, it shows how to locate objects in murky waters quickly and accurately. Overall, this study offers information on how to create underwater object identification algorithms that are optimized, which can increase the effectiveness and efficiency of ROV systems and help to lessen the environmental impact of marine garbage or help in the research of the marine environment.
