VALSE 论文速览 第100期：基于DETR框架和 Hard-Positive Query挖掘的人与物体交互检测 ...

论文题目：Towards Hard-Positive Query Mining for DETR-based Human-Object Interaction Detection

作者列表：钟旭彬 (华南理工大学)、*丁⻓兴 (华南理工大学)、黎子建 (华南理工大学)、⻩少立 (腾讯AI Lab)

B站观看网址：

https://www.bilibili.com/video/BV1Bt4y1c74J/

论文摘要：

Human-Object Interaction (HOI)detection is a core task for high-level image understanding. Recently, Detection Transformer (DETR)- based HOI detectors have become popular due to their superior performance and efficient structure. However, these approaches typically adopt fixed HOI queries for all testing images, which is vulnerable to the location change of objects in one specific image. Accordingly, in this paper, we propose to enhance DETR's robustness by mining hard-positive queries, which are forced to make correct predictions using partial visual cues. First, we explicitly compose hard-positive queries according to the ground-truth (GT)position of labeled human-object pairs for each training image. Specifically, we shift the GT bounding boxes of each labeled human-object pair so that the shifted boxes cover only a certain portion of the GT ones. We encode the coordinates of the shifted boxes for each labeled human-object pair into an HOI query. Second, we implicitly construct another set of hard-positive queries by masking the top scores in cross-attention maps of the decoder layers. The masked attention maps then only cover partial important cues for HOI predictions. Finally, an alternate strategy is proposed that efficiently combines both types of hard queries. In each iteration, both DETR's learnable queries and one selected type of hard- positive queries are adopted for loss computation. Experimental results show that our proposed approach can be widely applied to existing DETR-based HOI detectors. Moreover, we consistently achieve state-of-the-art performance on three benchmarks: HICO- DET, V-COCO, and HOI-A.

论文信息：

[1] X. Zhong, C. Ding, Z. Li, and S. Huang. Towards Hard-Positive Query Mining for DETR-based Human-Object Interaction Detection. In ECCV, 2022.

论文链接：

[https://arxiv.org/pdf/2207.05293.pdf]

代码链接：

[https://github.com/MuchHair/HQM]

视频讲者简介：

钟旭彬，华南理工大学博士生。主要研究方向包括视觉关系检测，人与物体交互检测 等。在IJCV、CVPR和ECCV等顶级期刊和会议发表多篇论文。

特别鸣谢本次论文速览主要组织者：

月度轮值AC：王立君 (大连理工大学)、眭亚楠 (清华大学)

季度责任AC：张姗姗 (南京理工大学)