Automatic recognition, counting and measurement of fish from underwater stereo-video

Underwater stereo-video systems are used for counting and measuring fish in aquaculture, fisheries and conservation management. Decreases in the costs and size of video cameras and increases in image resolution have made this... [ view full abstract ]

Underwater stereo-video systems are used for counting and measuring fish in aquaculture, fisheries and conservation management. Decreases in the costs and size of video cameras and increases in image resolution have made this technology more accessible. There has also been a rapid uptake of video based sampling in the marine environment over the last 2 decades.

The manual identification, counting and measuring of fish in video images by human operators is labour intensive, costly and is a major disincentive against the uptake of this technology. Automating species identification, counts and length measurement using computer vision and machine learning techniques will have a huge impact on marine science and conservation and fisheries management. We present results for both identification and measurement and discuss future directions.