Learning Rotation-Invariant Convolutional Neural Networks for Object Detection in VHR Optical Remote Sensing Images

Short summary

A novel Rotation-Invariant Convolutional Neural Network (RICNN) is proposed for object detection in very high resolution optical remote sensing images, achieving rotation invariance by learning a new layer with a regularization constraint that maps features of rotated samples close to each other.

AI-generated from the title and abstract; the full text is not read.