Region-Based Convolutional Networks for Accurate Object Detection and Segmentation

Ross Girshick, Jeff Donahue, Trevor Darrell, Jitendra Malik

IEEE Transactions on Pattern Analysis and Machine Intelligence · 2015 · 2.9K citations · 63 references

Concepts

TL;DR

Detection performance on the PASCAL VOC Challenge plateaued, and the best methods were complex ensembles that combined low‑level image features with high‑level context. The authors propose a simple and scalable detection algorithm. They combine high‑capacity convolutional networks with bottom‑up region proposals and employ supervised pre‑training on an auxiliary task followed by domain‑specific fine‑tuning, resulting in the R‑CNN model. The algorithm improves mean average precision by more than 50 % over the previous best on VOC 2012, achieving 62.4 %, and its source code is available at http://www.cs.berkeley.edu/~rbg/rcnn.

Abstract

Object detection performance, as measured on the canonical PASCAL VOC Challenge datasets, plateaued in the final years of the competition. The best-performing methods were complex ensemble systems that typically combined 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 50 percent relative to the previous best result on VOC 2012-achieving a mAP of 62.4 percent. Our approach combines two ideas: (1) one can apply high-capacity convolutional networks (CNNs) to bottom-up region proposals in order to localize and segment objects and (2) when labeled training data are scarce, supervised pre-training for an auxiliary task, followed by domain-specific fine-tuning, boosts performance significantly. Since we combine region proposals with CNNs, we call the resulting model an R-CNN or Region-based Convolutional Network. Source code for the complete system is available at http://www.cs.berkeley.edu/~rbg/rcnn.

References

63