Environmentally adaptive fish or no-fish classification for river video fish counters using high-performance desktop and embedded hardware

Download PDF
Open Webpage

Jürgen Soom, Vishwajeet Pattanaik, Mairo Leier, Jeffrey A. Tuhtan

Published: 01 Jan 2022, Last Modified: 07 Nov 2025Ecol. Informatics 2022EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•Six environmental condition classes are used as fish classification model parameters.•Best performance: 88.7% fish or no-fish classification accuracy, F1 score of 89.1%.•18–45 fps on embedded hardware and up to 150 fps using a high-performance desktop.