arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2108.11501cs.CV
收藏

减少特征纠缠提高目标检测和属性识别性能

Improving Object Detection and Attribute Recognition by Feature Entanglement Reduction

University of Southern California(南加州大学)

Zhaoheng Zheng, Arka Sadhu, Ram Nevatia

摘要:

我们探索目标检测与两个属性:颜色和材料。该任务旨在同时检测物体并推断其颜色和材料。一种直接的方法是在通常的目标检测管道的最后添加属性头。然而,我们观察到这两个目标是冲突的:目标检测应该是属性无关的,属性在很大程度上是对象无关的。由标准检测网络计算的特征纠缠类别和属性特征;我们通过使用双流模型来解开它们,其中类别和属性特征是独立计算的,但分类头共享感兴趣区域(ROI)。与传统的单流模型相比,我们的模型在监督和属性转移任务上都比VG-20(Visual Genome的一个子集)有显着的改进。

英文摘要:

We explore object detection with two attributes: color and material. The task aims to simultaneously detect objects and infer their color and material. A straight-forward approach is to add attribute heads at the very end of a usual object detection pipeline. However, we observe that the two goals are in conflict: Object detection should be attribute-independent and attributes be largely object-independent. Features computed by a standard detection network entangle the category and attribute features; we disentangle them by the use of a two-stream model where the category and attribute features are computed independently but the classification heads share Regions of Interest (RoIs). Compared with a traditional single-stream model, our model shows significant improvements over VG-20, a subset of Visual Genome, on both supervised and attribute transfer tasks.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑