[论文解读] Occluded Prohibited Items Detection: an X-ray Security Inspection Benchmark and De-occlusion Attention Module
本文提出了 OPIXray,这是首个针对X光安检图像中遮挡违禁品检测的高质量基准数据集,重点关注不同遮挡程度下的剪刀类物品。本文提出了一种即插即用的注意力机制——去遮挡注意力模块(De-occlusion Attention Module, DOAM),该模块利用多尺度外观线索来增强特征表示,在 SSD、YOLOv3 和 FCOS 等多种先进检测器上显著提升了检测性能,尤其在高遮挡场景下表现突出。
Security inspection often deals with a piece of baggage or suitcase where objects are heavily overlapped with each other, resulting in an unsatisfactory performance for prohibited items detection in X-ray images. In the literature, there have been rare studies and datasets touching this important topic. In this work, we contribute the first high-quality object detection dataset for security inspection, named Occluded Prohibited Items X-ray (OPIXray) image benchmark. OPIXray focused on the widely-occurred prohibited item "cutter", annotated manually by professional inspectors from the international airport. The test set is further divided into three occlusion levels to better understand the performance of detectors. Furthermore, to deal with the occlusion in X-ray images detection, we propose the De-occlusion Attention Module (DOAM), a plug-and-play module that can be easily inserted into and thus promote most popular detectors. Despite the heavy occlusion in X-ray imaging, shape appearance of objects can be preserved well, and meanwhile different materials visually appear with different colors and textures. Motivated by these observations, our DOAM simultaneously leverages the different appearance information of the prohibited item to generate the attention map, which helps refine feature maps for the general detectors. We comprehensively evaluate our module on the OPIXray dataset, and demonstrate that our module can consistently improve the performance of the state-of-the-art detection methods such as SSD, FCOS, etc, and significantly outperforms several widely-used attention mechanisms. In particular, the advantages of DOAM are more significant in the scenarios with higher levels of occlusion, which demonstrates its potential application in real-world inspections. The OPIXray benchmark and our model are released at https://github.com/OPIXray-author/OPIXray.
研究动机与目标
- 解决 X 光安检检测中遮挡违禁品专用数据集缺乏的问题。
- 构建一个模拟真实行李遮挡情况的现实基准,涵盖多种遮挡水平。
- 提出一种即插即用的注意力模块(DOAM),以提升在严重遮挡条件下的检测性能。
- 验证 DOAM 在多种目标检测器上的有效性,并证明其优于现有注意力机制。
- 实现鲁棒检测模型在真实安检系统中的实际部署。
提出的方法
- 构建包含 8,885 张五类剪刀类物品 X 光图像的 OPIXray 基准数据集,并由专业机场安检人员进行标注。
- 设计去遮挡注意力模块(DOAM),联合利用不同材料外观的形状、纹理和颜色线索。
- 将 DOAM 作为轻量化、端到端可训练模块集成到现有目标检测器(如 SSD、FCOS、YOLOv3)中。
- 采用多尺度特征融合策略训练 DOAM,强调遮挡区域的边缘和区域级信息。
- 通过可学习的注意力图优化 DOAM,动态突出部分可见违禁品的显著特征。
- 在检测头之前应用 DOAM 对特征图进行优化,提升在遮挡条件下的定位与分类性能。
实验结果
研究问题
- RQ1在 X 光安检图像中,不同遮挡程度下,先进目标检测器的性能如何退化?
- RQ2即插即用的注意力模块能否在不改变网络结构的前提下有效提升遮挡违禁品的检测准确率?
- RQ3DOAM 机制是否在处理 X 光图像中的跨类别遮挡方面优于现有注意力模块?
- RQ4DOAM 的性能增益在不同遮挡水平和物体类别之间如何变化?
- RQ5DOAM 在 OPIXray 基准中对不同目标检测框架(如 SSD、FCOS、YOLOv3)的泛化能力如何?
主要发现
- OPIXray 基准包含 8,885 张 X 光图像,涵盖 5 种剪刀类物品,由专业安检人员标注,并包含三个测试集,遮挡程度逐级递增。
- DOAM 在 SSD 上提升 mAP 3.12%,在 YOLOv3 上提升 1.04%,在 FCOS 上提升 0.39%,在各类检测器中均表现出一致的性能增益。
- DOAM 的性能增益在高遮挡场景下最为显著,SSD 基线中 SC(剪刀)的 mAP 提升达 41.50%,UT(多功能刀)提升达 95.12%。
- 注意力可视化结果表明,即使物品部分被遮挡,DOAM 仍能有效突出物体边界和显著区域。
- 在遮挡程度较低的类别(如折叠刀、剪刀)上,DOAM 略微降低性能,可能是因为对高度遮挡区域的注意力过强。
- DOAM 模块作为即插即用组件效果显著,可在不修改主干网络或检测头结构的前提下增强特征表示。
更好的研究,从现在开始
从阅读论文到最终审阅,大幅缩短您的研究时间。
无需绑定信用卡
本解读由 AI 生成,并经人工编辑审核。