Automated Analysis of Fisheries Information from Digital Stills

Toyon proposes development of a system which performs automated analysis of images for fish population monitoring and fishing regulation enforcement applications. The proposed system is capable of processing images collected from aircraft, including unmanned aerial vehicles, as well as images collected using boat-mounted or handheld cameras used to observe fish catches landed on the decks of vessels. The proposed system provides automated fish classification using images of fish collected against a wide range of relevant backgrounds, both for individual fish and for groups of fish assembled in any configuration. Furthermore, the system provides fish size estimation using two or more digital still images collected from unknown ranges. The proposed system is based on advanced algorithms, which will be implemented in real-time software with a convenient user interface. In Phase I, data collection will be performed for species of fish found in the Pacific Ocean. Algorithms for fish species classification and fish size estimation will be developed, tested, and evaluated. A comprehensive feasibility study will be performed, and the results of the research and development will be documented. In addition, an initial design for the Phase II system will be presented, and recommendations for future research and development will be provided.