Pig weight identification method and system

The invention discloses a pig re-identification method and system, and the method comprises the following steps: S1, obtaining an image: obtaining a picture or a picture sequence from a picture shooting interface; s2, target detection: detecting a pig individual by using a target detection model, and intercepting a region of interest; s3, feature extraction: performing feature extraction on the intercepted region through a feature extraction network to obtain feature information; s4, metric learning is carried out, feature distances between the feature information and all individuals in the base library are calculated, and similarity is calculated according to the feature distances; and S5, test evaluation: sorting the candidate individuals according to the similarity, and selecting the individual with the highest similarity as a result output. Compared with traditional manual identification, the efficiency is improved by multiple orders of magnitude, the workload and risk of employees are greatly reduced, th