APPLICATION OF A NEURAL NETWORK TO PREDICT THE PATH OF OBJECTS TRACKED IN A VIDEO SEQUENCE
S.V. Savelyev1,2, V.P. Kosykh1
1Institute of Automation and Electrometry, Siberian Branch, Russian Academy of Sciences, Novosibirsk, Russia
2Novosibirsk State University, Novosibirsk, Russia
Keywords: multi-object tracking, tracking-by-detection, neural network, ByteTrack
Abstract
This paper explores the integration of computationally efficient neural networks into a multi-object tracking algorithm in video sequences to reliably predict subsequent trajectories. Several well-known neural network models are tested, and a new one is proposed. The models examined demonstrate improved trajectory prediction performance compared to the baseline ByteTrack algorithm, without significantly impacting performance.
|
