Focal Loss for Dense Object Detection
Authors:
- Tsung-Yi LinTLTsung-Yi Lin
- Priya GoyalPGPriya Goyal
- Ross GirshickRGRoss Girshick
- Kaiming HeKHKaiming He
- Piotr DollarPDPiotr Dollar
Top Cited Papers
- 1 October 2017
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 2999-3007
Abstract
The highest accuracy object detectors to date are based on a two-stage approach popularized by R-CNN, where a classifier is applied to a sparse set of candidate object locations. In contrast, one-stage detectors that are applied over a regular, dense sampling of possible object locations have the potential to be faster and simpler, but have trailed the accuracy of two-stage detectors thus far. In this paper, we investigate why this is the case. We discover that the extreme foreground-background class imbalance encountered during training of dense detectors is the central cause. We propose to address this class imbalance by reshaping the standard cross entropy loss such that it down-weights the loss assigned to well-classified examples. Our novel Focal Loss focuses training on a sparse set of hard examples and prevents the vast number of easy negatives from overwhelming the detector during training. To evaluate the effectiveness of our loss, we design and train a simple dense detector we call RetinaNet. Our results show that when trained with the focal loss, RetinaNet is able to match the speed of previous one-stage detectors while surpassing the accuracy of all existing state-of-the-art two-stage detectors.Keywords
All Related Versions
This publication has 21 references indexed in Scilit:
- Mask R-CNN2017
- Speed/Accuracy Trade-Offs for Modern Convolutional Object Detectors2017
- Deep Residual Learning for Image Recognition2016
- Training Region-Based Object Detectors with Online Hard Example Mining2016
- Fast R-CNN2015
- Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation2014
- Selective Search for Object RecognitionInternational Journal of Computer Vision, 2013
- Rapid object detection using a boosted cascade of simple features2005
- Original approach for the localisation of objects in imagesIEE Proceedings - Vision, Image, and Signal Processing, 1994
- Backpropagation Applied to Handwritten Zip Code RecognitionNeural Computation, 1989