A Part-based Deep Learning Network for identifying individual crabs using abdomen images

Crabs, such as swimming crabs and mud crabs, are famous for their high nutritional value but are difficult to preserve. Thus, the traceability of crabs is vital for food safety. Existing deep-learning methods can be applied to identify individual crabs. However, there is no previous study that used abdomen images to identify individual crabs. In this paper, we provide a novel Part-based Deep Learning Network (PDN) to reliably identify an individual crab from its abdomen images captured under various conditions. In our PDN, we developed three non-overlapping and three overlapping partitions strategies of the abdomen image and further designed a part attention block. A swimming crab (Crab-201) dataset with the abdomen images of 201 swimming crabs and a more complex mud crab dataset (Crab-146) were collected to train and test the proposed PDN. Experimental results show that the proposed PDN using the overlapping partition strategy is better than the non-overlapping partition strategy. The edge texture of the abdomen has more identifiable features than the sulciform texture of the lower part of the abdomen. It also demonstrates that the proposed PDNOS3, which emphasizes the edge texture of the abdomen with overlapping partition strategies, is more reliable and accurate than the counterpart methods to identify an individual crab.