Roto-translation Equivariant YOLO for Aerial Images

Benjamin Maurel, Samy Blusseau, Santiago Velasco-Forero, Teodora Petrisor

Submission Track: Extended Abstract

Keywords: Object Detection, CNN, YOLO, Equivariance, Symmetry

TL;DR: This work introduces an Equivariant One-Stage Object Detector incorporating group convolutions to handle rotational transformations.

Abstract: This work introduces Eq-YOLO, an Equivariant One-Stage Object Detector based on YOLO-v8 incorporating group convolutions to handle rotational transformations. We show the interest of using equivariant-transforms to improve the detection performance on rotated data over the regular YOLO-v8 model while dividing the number of parameters to train by a factor greater than three.

Submission Number: 33