Using AI To Automate Fish Passage Counts
A fish passage count is a vital metric for anyone concerned about wildlife management. It’s important for officials and dam operators to have an accurate count of the fish crossing a dam (via a fish ladder).
Currently, most fish passage counts are conducted by video:
- A video camera monitors a fish ladder
- As fish pass thru the ladder, they cross in front of the camera
- A trained fish counter reviews the video footage and tabulates the results
This is a tedious process, and often times sub-sampling is used to save time (with varying levels of accuracy). But this is the old way of doing things. We’ve invented a new way…
Boulder AI Counts Fish With Technology
Using our DNNCam system and proprietary software, Boulder AI has developed an automated system for counting fish passage. A machine vision camera system with an AI algorithm monitors the fish ladder. As a fish swims by, the camera system automatically categorizes and counts the fish.
- Our system can identify fish species with 96% accuracy
- Our system can measure fish length within 2cm of actual size
- Our system works even in turbid conditions (up to 50NTU in some cases)
The end result is a table of count data, along with size information, and even an actual photo of each fish (as desired).