Sampling Equivariant Self-attention Networks for Object Detection in Aerial Images
Object detection in aerial images has greater rotational and scale variations because of the overhead image capture, which requires the detection model to more flexibly handle geometric transformations.
BibTex:
