AUTHOR=Zhang Kejie , Xin Yu , Shi Ce , Xie Zhijun , Ren Zhiming 

TITLE=A Pyramidal Feature Fusion Model on Swimming Crab Portunus trituberculatus Re-identification

JOURNAL=Frontiers in Marine Science

VOLUME=Volume 9 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/marine-science/articles/10.3389/fmars.2022.845112

DOI=10.3389/fmars.2022.845112

ISSN=2296-7745

ABSTRACT=Swimming crab Portunus trituberculatus is a vital crab species in coastal areas of China. In this study, an individual Re-identification method, i.e, re-identification (ReID) technology for P. trituberculatus was proposed. This method takes the carapace texture of P. trituberculatus as 'biological fingerprint', and extracts carapace texture features to identify P. trituberculatus. In this paper, a re-identification method based on PFFM (Pyramidal Feature Fusion Model) on P. trituberculatus ReID was designed, which can extract global features and local features simultaneously. Furthermore, this method utilize WAM (Weight Adaptive Module) to improve ReID accuracy for the P. trituberculatus individuals with incomplete carapace. In order to strengthen the discrimination of the extracted features, this paper adopts triplet loss in model training process. The triplet loss can obviously improve the effectiveness of P. trituberculatus ReID. In the experiments, we collect 211 P. trituberculatus individuals as data set, and carry out the proposed PFFM on Effect of Pyramidal Model , P. trituberculatus Features Analysis and Comparisons to the State-of-the-Arts. The experimental results show that the mAP and Rank-1 values of our method reached 93.2% and 93% in the left half occlusion case., and mAP and Rank-1 values reached 71.8% and 75.4% in the upper half occlusion case. By the experiments the effectiveness and robustness of the proposed method is verified.