Fast and robust zebrafish segmentation and detection algorithm under different spectrum conditions

Abstract

Zebrafish is a vertebrate animal used for spectral neurobehavioural studies due to its robust endocrine system. Such studies generate large amounts of video data, making it too time-consuming and exhausting for a human observer to manually log their behaviour. Thus, computer vision techniques must be applied. Unfortunately, current commercial software for fish observation were developed for analysis under normal incandescent lighting and sunlight, thus they fail to work for spectral studies whereby the incident light spectrum is changed. This research develops a fast and robust algorithm to detect and segment fish under different lighting spectrum and benchmarks it against a commercial off-the-shelf software.