Multiple Drosophila Tracking System with Heading Direction

Abstract

Machine vision systems have been widely used for image analysis, especially that which is beyond human ability. In biology, studies of behavior help scientists to understand the relationship between sensory stimuli and animal responses. This typically requires the analysis and quantification of animal locomotion. In our work, we focus on the analysis of the locomotion of the fruit fly Drosophila melanogaster, a widely used model organism in biological research. Our system consists of two components: fly detection and tracking. Our system provides the ability to extract a group of flies as the objects of concern and furthermore determines the heading direction of each fly. As each fly moves, the system states are refined with a Kalman filter to obtain the optimal estimation. For the tracking step, combining information such as position and heading direction with assignment algorithms gives a successful tracking result. The use of heading direction increases the system efficiency when dealing with identity loss and flies swapping situations. The system can also operate with a variety of videos with different light intensities.

Peer-induced quiescence of male Drosophila melanogaster following copulation

“…The positions of male flies and their linear velocity were quantified using the fly tracking software TPro ( Sirigrivatanawong et al, 2017 ; Okuno et al, 2019 ) for a total time of 25 min after copulation. A correlation plot between inter-individual distance and velocity was calculated based on the two-dimensional kernel density estimation using R .…”

Visual Detection and Tracking of Pigs

“…Tracking is then solved using either the Hungarian algorithm or a greedy assignment. For hatched Drosophila flies, Sirigrivatanawong et al [SAT+17] have presented a very similar approach. After segmentation based on background subtraction, the individual flies are detected in the image and their orientation is determined.…”