A multi-target tracking platform for zebrafish based on deep neural network
Mingzhu Sun, Wensheng Li, Zihao Jiao, Xin Zhao · 2019
Zebrafish has been widely used in the field of biological behavior, for it is an excellent model organism. Many valuable biological data can be obtained by analyzing the behavioral characteristics of zebrafish. A large amount of data will be generated when studying the behavior of zebrafish. The current artificial statistical methods are inefficient and unpractical. Therefore, it is of great importance to get access to track the behavior of zebrafish automatically and accurately through software. We have proposed a zebrafish multi-target tracking algorithm based on deep convolution neural network to identify the location of multi-zebrafish with high tracking accuracy and speed. The tracking algorithm performs well in different videos with various numbers and sizes of zebrafish. It can greatly improve the efficiency of biological behavioral experiments.