Hide-and-Seek: A Data Augmentation Technique for Weakly-Supervised Localization and Beyond

Abstract

We propose 'Hide-and-Seek' a general purpose data augmentation technique, which is complementary to existing data augmentation techniques and is beneficial for various visual recognition tasks. The key idea is to hide patches in a training image randomly, in order to force the network to seek other relevant content when the most discriminative content is hidden. Our approach only needs to modify the input image and can work with any network to improve its performance. During testing, it does not need to hide any patches. The main advantage of Hide-and-Seek over existing data augmentation techniques is its ability to improve object localization accuracy in the weakly-supervised setting, and we therefore use this task to motivate the approach. However, Hide-and-Seek is not tied only to the image localization task, and can generalize to other forms of visual input like videos, as well as other recognition tasks like image classification, temporal action localization, semantic segmentation, emotion recognition, age/gender estimation, and person re-identification. We perform extensive experiments to showcase the advantage of Hide-and-Seek on these various visual recognition problems.

TPAMI submission. This is a journal extension of our ICCV 2017 paper arXiv:1704.04232

References in corpus (3)

• Convolutional Feature Masking for Joint Object and Stuff Segmentation
• Weakly Supervised Object Localization with Multi-fold Multiple Instance Learning
• Simultaneous Detection and Segmentation

Cited by in corpus (11)

• YOLOv4: Optimal Speed and Accuracy of Object Detection
• A Survey on Deep Semi-supervised Learning
• GridMask Data Augmentation
• GraphCrop: Subgraph Cropping for Graph Classification
• The nature of giant clumps in high-z discs: a deep-learning comparison of simulations and observations
• AID: Pushing the Performance Boundary of Human Pose Estimation with Information Dropping Augmentation
• FenceMask: A Data Augmentation Approach for Pre-extracted Image Features
• Evaluating COPY-BLEND Augmentation for Low Level Vision Tasks
• When Human Pose Estimation Meets Robustness: Adversarial Algorithms and Benchmarks
• A Multi-resolution Model for Histopathology Image Classification and Localization with Multiple Instance Learning
• PriorityCut: Occlusion-guided Regularization for Warp-based Image Animation