EFCPose: End-to-End Multi-Person Pose Estimation with Fully Convolutional Heads

CASIA OpenIR > 紫东太初大模型研究中心

	EFCPose: End-to-End Multi-Person Pose Estimation with Fully Convolutional Heads
	Wang Haixin1,2 ; Zhou Lu1 ; Chen Yingying1 ; Wang Jinqiao1,2,3,4
发表期刊	IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
	2023
页码	early access
摘要	Mainstream methods of multi-person pose estimation are not end-to-end. Recently, some methods build an end-to-end framework based on the DETR framework, aiming to eliminate the need for hand-crafted modules like heuristic grouping and NMS post-processing. However, these DETR-based methods suffer from a heavy memory burden of processing the high-resolution backbone feature maps with transformers. In this paper, we propose an end-to-end multi-person pose estimation method with a fully convolutional network, termed EFCPose. Different from DETR-based methods, it directly predicts instance-aware poses in a pixel-wise manner with lightweight convolutional heads, avoiding the heavy memory burden. Overall, our method adopts the center-offset formulation and a one-to-one label assignment strategy to achieve the multi-person pose estimation in an end-to-end manner. The main contribution of our fully convolutional heads includes two aspects. On the one hand, we propose an unaligned center-offset representation to learn more reliable semantic centers to replace the inconsistent geometric centers, improving the performance of instance detection. On the other hand, we propose a novel regression strategy named limb-aware adaptive regression, which leverages separate adaptive points to convert challenging long-range offsets into simplified short-range offsets and incorporates limb constraints to elevate the regression quality of joint offsets. Compared with current DETR-based end-to-end methods, EFCPose avoids high com putational complexity and achieves higher accuracy. Extensive experiments on COCO Keypoint and CrowdPose benchmarks show that EFCPose outperforms other state-of-the-art bottom up and single-stage methods without flipping augmentation.
收录类别	SCI
七大方向——子方向分类	图像视频处理与分析
国重实验室规划方向分类	视觉信息处理
是否有论文关联数据集需要存交	否
文献类型	期刊论文
条目标识符	http://ir.ia.ac.cn/handle/173211/57152
专题	紫东太初大模型研究中心
通讯作者	Wang Haixin; Chen Yingying
作者单位	1.Foundation Model Research Center, Institute of Automation, Chinese Academy of Sciences 2.School of Artificial Intelligence, University of Chinese Academy of Sciences 3.Wuhan AI Research 4.Peng Cheng Laboratory
第一作者单位	中国科学院自动化研究所
通讯作者单位	中国科学院自动化研究所
推荐引用方式 GB/T 7714	Wang Haixin,Zhou Lu,Chen Yingying,et al. EFCPose: End-to-End Multi-Person Pose Estimation with Fully Convolutional Heads[J]. IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY,2023:early access.
APA	Wang Haixin,Zhou Lu,Chen Yingying,&Wang Jinqiao.(2023).EFCPose: End-to-End Multi-Person Pose Estimation with Fully Convolutional Heads.IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY,early access.
MLA	Wang Haixin,et al."EFCPose: End-to-End Multi-Person Pose Estimation with Fully Convolutional Heads".IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY (2023):early access.

条目包含的文件		下载所有文件
文件名称/大小	文献类型	版本类型	开放类型	使用许可
EFCPose_End-to-End_M（4407KB）	期刊论文	作者接受稿	开放获取	CC BY-NC-SA	浏览下载