How good are detection proposals, really?
How good are detection proposals, really?
复制标题
DOI：
10.5244/c.28.24
复制
发表时间：
2014-06
期刊：
影响因子：
--
通讯作者：
J. Hosang;Rodrigo Benenson;B. Schiele
中科院分区：
文献类型：
--
作者：
J. Hosang;Rodrigo Benenson;B. Schiele
目前性能最好的Pascal VOC对象检测器采用检测建议来指导对象的搜索，从而避免了跨图像的穷举滑动窗口搜索。尽管检测建议很受欢迎，但在对象检测过程中使用它们时，尚不清楚做出了哪些权衡。我们对十种对象建议方法进行了深入分析，并提供了关于地面真相注释召回率（在Pascal VOC 2007和ImageNet 2013上）、可重复性和对DPM检测器性能的影响的四个基线。我们的研究结果显示了现有方法的共同弱点，并提供了针对不同设置选择最适当方法的见解。
Current top performing Pascal VOC object detectors employ detection proposals to guide the search for objects thereby avoiding exhaustive sliding window search across images. Despite the popularity of detection proposals, it is unclear which trade-offs are made when using them during object detection. We provide an in depth analysis of ten object proposal methods along with four baselines regarding ground truth annotation recall (on Pascal VOC 2007 and ImageNet 2013), repeatability, and impact on DPM detector performance. Our findings show common weaknesses of existing methods, and provide insights to choose the most adequate method for different settings.