Research / Imaging and Perception Systems for Robotics / Marine robotics and perception

FishEye: Robust 3D Fish Measurement from Single Image or Video

We’re developing a low-cost way to estimate a fish’s length and volume from a photo or video.

FishEye detects the fish, estimates depth, and reconstructs its body in 3D for measurement.
FishEye detects the fish, estimates depth, and reconstructs its body in 3D for measurement.

Project overview

FishEye uses deep learning to detect and segment fish in photos or videos from ordinary cameras, including smartphones. It uses transformer-based camera calibration and depth estimation to reconstruct each fish in 3D and estimate its length. A ping-pong ball can serve as a scale reference. We’re developing FishEye for field measurements in fisheries, ecological research, and aquaculture.

Related project · FishPod

Project team

Project lead
Ethan Chung

Principal investigator
Huaijin Chen

Contributors
Vincent Chan