Paper
Rich feature hierarchies for accurate object detection and semantic segmentation
Authors
Ross Girshick, Jeff Donahue, Trevor Darrell, Jitendra Malik
Abstract
Object detection performance, as measured on the canonical PASCAL VOC
dataset, has plateaued in the last few years. The best-performing methods are
complex ensemble systems that typically combine multiple low-level image
features with high-level context. In this paper, we propose a simple and
scalable detection algorithm that improves mean average precision (mAP) by more
than 30% relative to the previous best result on VOC 2012---achieving a mAP of
53.3%. Our approach combines two key insights: (1) one can apply high-capacity
convolutional neural networks (CNNs) to bottom-up region proposals in order to
localize and segment objects and (2) when labeled training data is scarce,
supervised pre-training for an auxiliary task, followed by domain-specific
fine-tuning, yields a significant performance boost. Since we combine region
proposals with CNNs, we call our method R-CNN: Regions with CNN features. We
also compare R-CNN to OverFeat, a recently proposed sliding-window detector
based on a similar CNN architecture. We find that R-CNN outperforms OverFeat by
a large margin on the 200-class ILSVRC2013 detection dataset. Source code for
the complete system is available at http://www.cs.berkeley.edu/~rbg/rcnn.