Oriented Response Networks

Abstract

Deep Convolution Neural Networks (DCNNs) are capable of learning unprecedentedly effective image representations. However, their ability in handling significant local and global image rotations remains limited. In this paper, we propose Active Rotating Filters (ARFs) that actively rotate during convolution and produce feature maps with location and orientation explicitly encoded. An ARF acts as a virtual filter bank containing the filter itself and its multiple unmaterialised rotated versions. During back-propagation, an ARF is collectively updated using errors from all its rotated versions. DCNNs using ARFs, referred to as Oriented Response Networks (ORNs), can produce within-class rotation-invariant deep features while maintaining inter-class discrimination for classification tasks. The oriented response produced by ORNs can also be used for image and object orientation estimation tasks. Over multiple state-of-the-art DCNN architectures, such as VGG, ResNet, and STN, we consistently observe that replacing regular filters with the proposed ARFs leads to significant reduction in the number of network parameters and improvement in classification performance. We report the best results on several commonly used benchmarks.

arXiv commentAccepted in CVPR 2017. Source code available at http://yzhou.work/ORN

References in corpus (11)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
• ADADELTA: An Adaptive Learning Rate Method
• Deep Residual Learning for Image Recognition
• Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)
• Wide Residual Networks
• Highway Networks
• Learning Deep Features for Discriminative Localization
• Convolutional Oriented Boundaries
• Generalizing Pooling Functions in Convolutional Neural Networks: Mixed, Gated, and Tree
• Oriented Edge Forests for Boundary Detection

Cited by in corpus (20)

• Gabor Convolutional Networks
• Deep Learning for UAV-based Object Detection and Tracking: A Survey
• Land cover mapping at very high resolution with rotation equivariant CNNs: towards small yet accurate models
• Rotation equivariant vector field networks
• Deep Learning for Generic Object Detection: A Survey
• Learning a Rotation Invariant Detector with Rotatable Bounding Box
• Scene Text Detection and Recognition: The Deep Learning Era
• What Do We Understand About Convolutional Networks?
• Scale equivariance in CNNs with vector fields
• Learning Invariances in Neural Networks
• Rotation-Sensitive Regression for Oriented Scene Text Detection
• Projection Convolutional Neural Networks for 1-bit CNNs via Discrete Back Propagation
• Robustness of Rotation-Equivariant Networks to Adversarial Perturbations
• B-Spline CNNs on Lie Groups
• RotDCF: Decomposition of Convolutional Filters for Rotation-Equivariant Deep Networks
• Towards a Better Match in Siamese Network Based Visual Object Tracker
• Affine Self Convolution
• Integrating Multiple Receptive Fields through Grouped Active Convolution
• Rotational Rectification Network: Enabling Pedestrian Detection for Mobile Vision
• A Compositional Textual Model for Recognition of Imperfect Word Images