Open AccessHow good are detection proposals, really?
Author(s) -
Jan Hosang,
Rodrigo Benenson,
Bernt Schiele
Publication year - 2014
Language(s) - English
Resource type - Conference proceedings
DOI - 10.5244/c.28.24
Subject(s) - pascal (unit) , computer science , object detection , popularity , detector , recall , sliding window protocol , artificial intelligence , ground truth , precision and recall , strengths and weaknesses , data mining , window (computing) , pattern recognition (psychology) , psychology , social psychology , telecommunications , cognitive psychology , programming language , operating system
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.
Accelerating Research
Robert Robinson Avenue,
Oxford Science Park, Oxford
OX4 4GP, United Kingdom
Address
John Eccles HouseRobert Robinson Avenue,
Oxford Science Park, Oxford
OX4 4GP, United Kingdom