Unified Human-Centric Model, Framework and Benchmark: A Survey

Human-centric Computer Vision Tasks (HCTs) refer to a series of tasks related to the human body, such as Human Pose Estimation, Pedestrian Tracking, Re-Identification (ReID), Human Parsing and Action Recognition, etc. In the past three years, a large number of Human-centric Methods (HCMds) for HCTs...

Full description

Saved in:
Bibliographic Details
Published in:IEEE access 2024, Vol.12, p.155408-155422
Main Authors: Zhao, Xiong, Sulaiman, Sarina, Yee Leng, Wong
Format: Article
Language:English
Subjects:
Benchmark testing
Benchmarks
Computational modeling
Computer vision
Data models
Human body
Human computer interaction
Human factors
Human-centric
Literature reviews
Multi-modal
Multimodal sensors
Pose estimation
Surveys
Task analysis
Taxonomy
Tokenization
tokenize
unified model
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Human-centric Computer Vision Tasks (HCTs) refer to a series of tasks related to the human body, such as Human Pose Estimation, Pedestrian Tracking, Re-Identification (ReID), Human Parsing and Action Recognition, etc. In the past three years, a large number of Human-centric Methods (HCMds) for HCTs have emerged, based on a common assumption that these tasks should share the same underlying semantic structure of the human body, integrating multi-modal information could achieve more powerful functionality or lower computational cost. However, a systematic and comprehensive literature review on this field is still missing, this survey provides a comprehensive review of these works. We first give a clear definition and taxonomy standard, then propose a new taxonomy of HCMds. Next, we discuss key technologies and the path towards Unified Human-centric model (UniHCM). Third, following the new taxonomy, we take a brief review and summary highlights of the representative model, framework, and benchmark for HCTs. Finally, We discuss and analyze the limitations of existing HCMds and suggest possible future research directions.
ISSN:2169-3536
2169-3536
DOI:10.1109/ACCESS.2024.3450123