Identification and Measurement of Sea Cucumber Based on YOLOv7 and GrabCut-RGBD

The sea cucumber fishing ROV needs to select and grade the sea cucumber when it is fishing the sea cucumber. However, refraction and image quality seriously affect the recognition of sea cucumber by sea cucumber fishing ROV and the accurate measurement of sea cucumber size. This paper presents a new method of sea cucumber identification and size measurement based on YOLOv7 and GrabCut-RGBD. On the left eye correction image, the pretrained YOLOv7 sea cucumber detection model was used to identify the sea cucumber and locate the region of interest. A GrabCut-RGBD image segmentation method was used to segment two-dimensional sea cucumber objects on the region of interest, so as to obtain two-dimensional sea cucumber dimensions. The three-dimensional coordinates are obtained by triangulation, and the optimal size measurement point is found on the target image of sea cucumber to realize automatic size measurement of sea cucumber. The experimental results show that the accuracy of sea cucumber identification is 97%, and the average error of sea cucumber size measurement is 6%, which is better than the traditional method in sea cucumber identification and size measurement.