Abstract:In recent years, convolutional neural network has shown good performance in many image processing and computer vision tasks. However, a standard CNN model is not invariant to image rotations. In fact, even slight rotation of an input image will seriously degrade its performance. This shortcoming precludes the use of CNN in some practical scenarios. Thus, in this paper, we focus on designing convolutional layer with good rotation invariance. Specifically, based on a simple rotation-invariant coordinate system, we propose a new convolutional operation, called Rotation-Invariant Coordinate Convolution (RIC-C). Without additional trainable parameters and data augmentation, RIC-C is naturally invariant to arbitrary rotations around the input center. Furthermore, we find the connection between RIC-C and deformable convolution, and propose a simple but efficient approach to implement RIC-C using Pytorch. By replacing all standard convolutional layers in a CNN with the corresponding RIC-C, a RIC-CNN can be derived. Using MNIST dataset, we first evaluate the rotation invariance of RIC-CNN and compare its performance with most of existing rotation-invariant CNN models. It can be observed that RIC-CNN achieves the state-of-the-art classification on the rotated test dataset of MNIST. Then, we deploy RIC-C to VGG, ResNet and DenseNet, and conduct the classification experiments on two real image datasets. Also, a shallow CNN and the corresponding RIC-CNN are trained to extract image patch descriptors, and we compare their performance in patch verification. These experimental results again show that RIC-C can be easily used as drop in replacement for standard convolutions, and greatly enhances the rotation invariance of CNN models designed for different applications.

Sorting Convolution Operation for Achieving Rotational Invariance

Achieving Rotation Invariance in Convolution Operations: Shifting from Data-Driven to Mechanism-Assured

Polar Coordinate Convolutional Neural Network: from Rotation-Invariance to Translation-Invariance

RIC-CNN: Rotation-Invariant Coordinate Convolutional Neural Network

Gradient-Aligned convolution neural network

Rotation Invariant Local Binary Convolution Neural Networks.

Rotation Equivariance and Invariance in Convolutional Neural Networks

Rotation-Invariant Feature Learning Via Convolutional Neural Network with Cyclic Polar Coordinates Convolutional Layer

Patch Reordering: a Novel Way to Achieve Rotation and Translation Invariance in Convolutional Neural Networks

Transform-Invariant Convolutional Neural Networks for Image Classification and Search

Invariant Feature Extraction for Image Classification Via Multi-Channel Convolutional Neural Network

RC-CNN: Representation-Consistent Convolutional Neural Networks for Achieving Transformation Invariance

Compensating for Large In-Plane Rotations in Natural Images

Equivariance-bridged SO(2)-Invariant Representation Learning using Graph Convolutional Network

A rotation meanout network with invariance for dermoscopy image classification and retrieval

RIConv++: Effective Rotation Invariant Convolutions for 3D Point Clouds Deep Learning

RRL:Regional Rotation Layer in Convolutional Neural Networks

Rotational Objects Recognition and Angle Estimation Via Kernel-Mapping CNN

Empowering Networks With Scale and Rotation Equivariance Using A Similarity Convolution

Roto-translation equivariant convolutional networks: Application to histopathology image analysis

Rotation Invariant Gabor Convolutional Neural Network for Image Classification