Artificial Intelligence (AI) is revolutionizing ecology and conservation by enabling species recognition from photos and videos. Our project evaluates the capacity to expand AI for individual fish recognition for population assessment. The success of this effort would facilitate fisheries analysis at an unprecedented scale by engaging anglers and citizen scientists in imagery collection. This project is one of the first attempts to apply AI towards fish population assessment.
Our study seeks to apply technological advances in Artificial Intelligence (AI), machine learning, and high-performance computing to enable new methods for fish biology and population assessment. We propose 3 main research objectives: (1) assemble and annotate fish imagery for AI training data, (2) develop and evaluate CNN models for classification of species and individuals, (3) evaluate the capacity for AI application in riverine fish population assessment based on estimated error rates. We are using imagery of wild brook trout (Figure 1) and other species to train and test individual-recognition models. This modeling approach could enable anglers to contribute to quantitative fish population assessments based on imagery of fish they encounter. This project is supported by the USGS Community for Data Integration and the USGS Eastern Ecological Science Center.
Presentation:
Hitt, N.P., N. Rapstine, M. Arami, K. Kessler, H. Macmillan, and B. Letcher. 2020. Deep learning identifies population structure and individual variation in native brook trout. Federation of Earth Science Information Partners (ESIP), Summer Meeting. Contributed presentation.
Comparison of underwater video with electrofishing and dive‐counts for stream fish abundance estimation
- Overview
Artificial Intelligence (AI) is revolutionizing ecology and conservation by enabling species recognition from photos and videos. Our project evaluates the capacity to expand AI for individual fish recognition for population assessment. The success of this effort would facilitate fisheries analysis at an unprecedented scale by engaging anglers and citizen scientists in imagery collection. This project is one of the first attempts to apply AI towards fish population assessment.
Our study seeks to apply technological advances in Artificial Intelligence (AI), machine learning, and high-performance computing to enable new methods for fish biology and population assessment. We propose 3 main research objectives: (1) assemble and annotate fish imagery for AI training data, (2) develop and evaluate CNN models for classification of species and individuals, (3) evaluate the capacity for AI application in riverine fish population assessment based on estimated error rates. We are using imagery of wild brook trout (Figure 1) and other species to train and test individual-recognition models. This modeling approach could enable anglers to contribute to quantitative fish population assessments based on imagery of fish they encounter. This project is supported by the USGS Community for Data Integration and the USGS Eastern Ecological Science Center.
Presentation:
Hitt, N.P., N. Rapstine, M. Arami, K. Kessler, H. Macmillan, and B. Letcher. 2020. Deep learning identifies population structure and individual variation in native brook trout. Federation of Earth Science Information Partners (ESIP), Summer Meeting. Contributed presentation.
- Publications
Comparison of underwater video with electrofishing and dive‐counts for stream fish abundance estimation
Advances in video technology enable new strategies for stream fish research. We compared juvenile (age‐0) and adult (age 1+) Brook Trout Salvelinus fontinalis abundance estimates from underwater video with backpack electrofishing and dive‐count methods across a series of stream pools in Shenandoah National Park, Virginia (n = 41). Video methods estimated greater mean abundance of adult trout than