Target Tracking Analysis of Physical Exercise Video Based on Object Detection Algorithm

This paper aims at the current application status and challenges of target detection and tracking technology in the field of physical exercise video analysis, and clearly points out the important value of accurate and real-time target positioning and track tracking of sports individuals for improvin...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:2024 2nd International Conference on Mechatronics, IoT and Industrial Informatics (ICMIII) s. 343 - 347
Hlavní autor: Xu, Chunmiao
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 12.06.2024
Témata:
Accuracy
Kalman filtering
Object detection
Object detection algorithm
Statistical analysis
Streaming media
Target tracking
Target tracking analysis
Training
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:This paper aims at the current application status and challenges of target detection and tracking technology in the field of physical exercise video analysis, and clearly points out the important value of accurate and real-time target positioning and track tracking of sports individuals for improving physical education, training effect evaluation and athlete movement analysis. Advanced deep learning object detection model is used to identify and locate the subject object in the physical exercise video efficiently and accurately. On this basis, combined with Kalman filter prediction model, continuous tracking of moving targets is carried out to deal with the problem of target loss in complex cases such as fast motion, occlusion and attitude change. Through the experimental verification of the actual physical exercise video data set, the method can achieve stable target tracking in various sports scenes, effectively extract the exercisers' movement track, speed, direction and other key information, and then conduct quantitative analysis of the effectiveness, standardization and physical energy consumption of exercise actions in multiple dimensions. The research results will provide strong technical support for the fields of intelligent physical education, sports science research and remote fitness guidance.
DOI:10.1109/ICMIII62623.2024.00069