Unsupervised Hard Example Mining from Videos for Improved Object Detection
arXiv:1808.04285
Abstract
Important gains have recently been obtained in object detection by using training objectives that focus on {\em hard negative} examples, i.e., negative examples that are currently rated as positive or ambiguous by the detector. These examples can strongly influence parameters when the network is trained to correct them. Unfortunately, they are often sparse in the training data, and are expensive to obtain. In this work, we show how large numbers of hard negatives can be obtained {\em automatically} by analyzing the output of a trained detector on video sequences. In particular, detections that are {\em isolated in time}, i.e., that have no associated preceding or following detections, are likely to be hard negatives. We describe simple procedures for mining large numbers of such hard negatives (and also hard {\em positives}) from unlabeled video data. Our experiments show that retraining detectors on these automatically obtained examples often significantly improves performance. We present experiments on multiple architectures and multiple data sets, including face detection, pedestrian detection and other object categories.
arXiv comment14 pages, 7 figures, accepted at ECCV 2018
References in corpus (16)
- Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
- Focal Loss for Dense Object Detection
- Synthesizing Robust Adversarial Examples
- Online Batch Selection for Faster Training of Neural Networks
- Foveation-based Mechanisms Alleviate Adversarial Examples
- From Facial Parts Responses to Face Detection: A Deep Learning Approach
- Detecting Faces Using Region-based Fully Convolutional Networks
- How Far are We from Solving Pedestrian Detection?
- Illuminating Pedestrians via Simultaneous Detection & Segmentation
- Fracking Deep Convolutional Image Descriptors
- Learning Complexity-Aware Cascades for Deep Pedestrian Detection
- Taking a Deeper Look at Pedestrians
- Bootstrapping Face Detection with Hard Negative Examples
- SFD: Single Shot Scale-invariant Face Detector
- Face Detection using Deep Learning: An Improved Faster RCNN Approach
- Fine-tuning deep CNN models on specific MS COCO categories