Smeltcam: System for counting and identifying fish.

Smeltcam is a specialized form of a general tool developed by Sureworks that can detect fish at a depth of 30ft in turbid water, count and recognize species using advanced computer vision techniques. The device is attached to a trawling net and towed. Fish pass safely through the device using infrared light. As objects pass through, their pictures are taken at 60 frames per second and analyzed. The system has also been deployed in a fish ladder environment to perform the same functions, but in a stationary environment.

Here is a link to the paper published in the Plos One journal on its use in the Sacramento Delta.




Below is a photo of the endangered delta smelt in the murky waters of the Sacramento delta:



delta smelt(4)


Device being deployed in the Sacramento Delta:IMG_20121126_150129-1


Thanks to a contract from the California State and Federal Contractors Water Agency, the newest form of the device shown below will measure several water quality parameters, depth, GPS and weigh approximately 1/3 as much as the existing prototype. In addition, thanks to a contract from the Metropolitan Water District of Southern California, it will also utilize Deep Learning, an advanced multi-layered neural net approach to species identification.



Smeltcam Rendering


Update (July 30, 2015): 

20150722_154534 (1)

Update (April, 2016): New unit produced for USBR in Sacramento. Has the ability to connect to any YSI EXO water quality device, improved weight, improved species recognition accuracy, improved data collection and user interface. Currently working with USBR to integrate the Teledyne RD ADCP into the system (a type of sonar that measures and records water current velocities over a range of depths).