Online Multiple Athlete Tracking with Pose-Based Long-Term Temporal Dependencies

Autor: Mengxiao Zhu, Rui He, Qingjie Liu, Nan Ran, Longteng Kong
Jazyk: angličtina
Rok vydání: 2021
Předmět:
Multi-Athlete Tracking (MAT)
Memory
Long-Term

Computer science
long short-term memory (LSTM) networks
ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION
02 engineering and technology
010501 environmental sciences
Machine learning
computer.software_genre
Tracking (particle physics)
lcsh:Chemical technology
01 natural sciences
Biochemistry
Article
Motion (physics)
Analytical Chemistry
Task (project management)
Motion
Similarity (psychology)
0202 electrical engineering
electronic engineering
information engineering

Humans
lcsh:TP1-1185
Electrical and Electronic Engineering
Instrumentation
Blossom algorithm
0105 earth and related environmental sciences
sports video analysis
business.industry
Atomic and Molecular Physics
and Optics

Term (temporal)
Athletes
020201 artificial intelligence & image processing
Artificial intelligence
business
computer
Algorithms
Sports
Zdroj: Sensors, Vol 21, Iss 197, p 197 (2021)
Sensors (Basel, Switzerland)
Sensors
Volume 21
Issue 1
ISSN: 1424-8220
Popis: This paper addresses the Multi-Athlete Tracking (MAT) problem, which plays a crucial role in sports video analysis. There exist specific challenges in MAT, e.g., athletes share a high similarity in appearance and frequently occlude with each other, making existing approaches not applicable for this task. To address this problem, we propose a novel online multiple athlete tracking approach which make use of long-term temporal pose dynamics for better distinguishing different athletes. Firstly, we design a Pose-based Triple Stream Network (PTSN) based on Long Short-Term Memory (LSTM) networks, capable of modeling long-term temporal pose dynamics of athletes, including pose-based appearance, motion and athletes&rsquo
interaction clues. Secondly, we propose a multi-state online matching algorithm based on bipartite graph matching and similarity scores produced by PTSN. It is robust to noisy detections and occlusions due to the reliable transitions of multiple detection states. We evaluate our method on the APIDIS, NCAA Basketball and VolleyTrack databases, and the experiment results demonstrate its effectiveness.
Databáze: OpenAIRE