Project B.1.a - Improving smart species identification tools
Program(s) in charge: Data Collection & Database Program - Ecosystem & Bycatch Program
Click to see projects by theme, goal or target:
- Develop smart tools for accurately identifying prioritized species
- Researchers of Michigan State University, Texas A&M University, and St. Anselm College have been funded by the National Science Foundation to develop smart tools for identifying species in diverse fisheries contexts.
- Tools under development consist of: i) a smartphone application that employs artificial intelligence (AI) to perform species identification using user-supplied photos or video, and ii) genomic tests to perform genetic species identification in the field.
- Together, these tools could make rapid and highly accurate species identification possible without the need for specialized training or equipment.
- Due to a variety of reasons, accurate species identification in the field (i.e., landing sites) or by observers or cameras on-board (e.g., purse-seines, longlines) is not always possible.
- Therefore, tools that improve species identification of prioritized species in a rapid and accurate manner are desirable.
- Relevance for management
- Improved species identification during data collection programs will increase data quality provision to enhance stock assessments and other biological and ecological studies for prioritized species performed by the IATTC staff, reducing uncertainty in the scientific-advice and decision making.
- A trained AI model could increase the effectiveness of algorithms to review records collected by Electronic Monitoring (EM) equipment in a rapid and accurate manner, and help implement EM-programs in the region.
- 24 months
- Workplan and status
- Year 1: Sampling and collection of tissue, photo and video collection of prioritized species by technicians in the field and on-board observers or EM-cameras to improve genetic analysis and the training of the AI model, respectively.
- Year 2: Beta testing of smartphone application and rapid genetic tests.
- These activities will require the collaboration of national authorities and fishing industry.
- External collaborators
- Michigan State University, Texas A&M University, and St. Anselm College, fishing industry, CPCs
- Improved smartphone application that employs an AI model to perform species identification using user-supplied photo or video.
- Improved genomic tests to perform genetic species identification in the field.
- Improved AI algorithm to review EM data in a rapid and accurate manner.
- Dissemination material (e.g., reports, presentations) for the Bycatch Working Group, the SAC, the Tuna Conference, and other meetings of interest.
- Updated date: 01 May 2023
- Progress summary for the reporting period
- A beta version of the smartphone app is currently being finalized and will be tested by IATTC observers beginning in spring-summer 2023.
Tissue sampling, footage storage and tagging protocols have been compiled and consolidated to match IATTC’s existing methods.
Sampling kits are being prepared and will be ready for IATTC observers beginning in spring-summer 2023. IATTC’s Central American Shark Programs’ existing footage has been reviewed, processed by species, and shared with collaborators.
The IATTC staff coordinated and shared images to support the development of a field guide for the identification of mobulid rays captured in Pacific Ocean fisheries. Staff also developed a photo library to assist with the coordination of footage provided to the iCatch program
IATTC staff is translating the Mobulid field guide into Spanish to support IATTC observers. Several guides will be printed and the guide will also be made available electronically to support CPCs and their observer programs with training materials.
The IATTC staff provided support to collaborators for funding applications.
- Challenges and key lessons learnt
- Obtaining a significant amount of species-specific footage is difficult, especially for rare bycatch species.