Automated molting detection system for commercial soft-shell crab (Portunus pelagicus) production

Cost, availability and reliability of labor are major problems in any aquaculture operation, and particularly in the soft-shell crab industry, since its success depends on the precise timing and accuracy of the workers to observe and harvest the newly molted crabs before their shells harden. To improve efficiency and reduce the dependence on human intervention, we have developed an automated molting detection system. The detection system utilizes the fact that the crab???s carapace reflects infrared light (appearing as white pixels) much more strongly than the surrounding areas. By using Internet Protocol (IP) cameras, network video recorder (NVR), personal computer and newly designed image analysis software, molting can be detected by continuously measuring relative changes in white pixel area. Two-dimensional Gaussian function and Otsu???s thresholding method are incorporated into the detection software. Snapshot images, date, and time of molting are recorded automatically and displayed via user interface. Test results indicated that the highest (100%) hit rate and lowest precision (13.92%) were obtained when detection threshold was 20%. Lower hit rates and higher precision were observed at higher threshold levels. The optimum threshold for detecting molting in commercial operations is discussed.

Automated summary

This article introduces an automated system for detecting the molting of soft-shell crabs, which uses relative changes in white pixel area to detect molting and records relevant information. Test results show that the highest hit rate for molting was achieved at a 20% threshold, but with lower precision.