DEEP LEARNING BASED SPECIES CLASSIFICATION OF CRABS USING EFFICIENT NET AND MOBILENETV2

The proposed system successfully demonstrates an automated approach for crab species classification using deep learning techniques. By combining MobileNetV2 and EfficientNet, the model effectively extracts both lightweight and fine grained features, improving overall classification performance. The integration of a hybrid loss function (ArcFace + Center Loss) enhances feature discrimination by ensuring better separation between different species and tighter clustering within the same species.The use of 128-dimensional feature embeddings along with cosine similarity enables efficient and scalable classification, even for visually similar crab species. The model achieves high accuracy of around 96%, indicating strong performance and reliability. Additionally, the system shows good generalization by handling variations in lighting, background, and image conditions.Overall, the proposed model provides a fast, accurate, and scalable solution for crab species identification. It can be effectively applied in real-world scenarios such as fisheries management, aquaculture, and biodiversity conservation, reducing dependency on manual identification and expert knowledge.

               [1]Feature Extraction and Recognition of Chinese Mitten Crab Carapace Based on Improved MobileNetV2-2024              [2]Song, X.; Huang, F.; Liang, F.; Hu, J.; Lu, B.; Cai, C. Identification of Geographical Origin and Culture Water Body of Chinese Mitten Handed Crab Eriocheir sinensis Based on Mineral Element Fingerprints. Chin. J. Fish- 2024              [3]Zhou, L.; Gao, J.; Yang, Y.; Nie, Z.; Liu, K.; Xu, G. Genetic Diversity and Population Structure Analysis of Chinese Mitten Crab (Eriocheir sinensis) in the Yangtze and Liaohe Rivers. Fishes -2023              [4]Xu, Y.; Xue, J.; Liu, H.; Jiang, T.; Chen, X.; Yang, J. Identification of “Bathed” Chinese Mitten Crabs (Eriocheir sinensis) Using Geometric Morphological Analysis of the Carapace. Fishes- 2023              [5]Minaee, S.; Abdolrashidi, A.; Su, H.; Bennamoun, M.; Zhang, D. Biometrics recognition using deep learning: A survey. Artif. Intell. Rev- 2023               [6]Valero-Carreras, D.; Alcaraz, J.; Landete, M. Comparing two SVM models through different metrics based on the confusion matrix. Comput. Oper. Res. 2023.              [7]Ahmad,M.;Abbas, S.; Fatima, A.; Issa, G.F.; Ghazal, T.M.; Khan, M.A. Deep transfer learning-based animal face identification model empowered with vision-based hybrid approach. Appl. Sci. 2023.              [8]Feng, Y.; Yang, X.; Xu, D.; Luo, N.; Chen, F.; Sun, C. Recognition method of individual images of river crabs based on transfer learning and pyramid convolutional network. Fish. Mod. 2022.              [9]Kim,J.H.; Poulose, A.; Colaco, S.J.; Neethirajan, S.; Han, D.S. Enhancing animal welfare with interaction recognition: A deep dive into pig interaction using xception architecture and SSPD-PIR method. Agriculture 2023, 13, 1522. [CrossRef]           [10]Wan, Z.; Tian, F.; Zhang, C. Sheep face recognition model based on deep learning and bilinear feature fusion. Animals 2023, 13, 1957. [CrossRef] [PubMed]