Academy & Industry Research Collaboration Center (AIRCC)

Volume 9, Number 17, December 2019

Augmentation for small object detection


Mate Kisantal1, Zbigniew Wojna1,2, Jakub Murawski2,3, Jacek Naruniec3 and Kyunghyun Cho4, 1Tensorflight, Inc., 2University College London, 3Warsaw University of Technology and 4New York University


In the recent years, object detection has experienced impressive progress. Despite these improvements, there is still a significant gap in the performance between the detection of small and large objects. We analyze the current state-of-the-art model, Mask-RCNN, on a challenging dataset, MS COCO. We show that the overlap between small ground-truth objects and the predicted anchors is much lower than the expected IoU threshold. We conjecture this is due to two factors; (1) only a few images are containing small objects, and (2) small objects do not appear enough even within each image containing them. We thus propose to oversample those images with small objects and augment each of those images by copy-pasting small objects many times. It allows us to trade off the quality of the detector on large objects with that on small objects. We evaluate different pasting augmentation strategies, and ultimately, we achieve 9.7% relative improvement on the instance segmentation and 7.1% on the object detection of small objects, compared to the current state of the art method on MS COCO.


object detection, small object detection, instance segmentation, small objects, data augmentation.