The invention provides a crab weight measuring
system and method based on vision, and the method comprises the steps: collecting an
Eriocheir sinensis image on a measuring platform through a fixed camera, extracting a crab shell
mask region based on a YOLOv8 image recognition technology, and calculating the pixel area of a target region; calculating a physical length corresponding to a unit pixel according to a reference object with a known size, and calculating an actual physical area of the crab
carapace; meanwhile, the pixel width and length of the crab
carapace are detected based on a key point detection
algorithm, the thickness influence is corrected through a
linear regression equation, the mapping relation between the pixel size and the actual physical size is established, and the actual
carapace size of the crab is calculated; and then an improved TabTransform model is provided, and the obtained morphological characteristics of the crabs are used as
data input, so that the weight of the crabs is measured. Therefore, the problems that in traditional
aquaculture, manual measurement is low in efficiency and prone to causing
stress injury to living bodies are solved, an automatic
body weight measurement mode is provided, and high-precision and high-efficiency
body weight measurement of crabs is achieved.