SWA Object Detection

TL;DR

This paper demonstrates that applying Stochastic Weights Averaging (SWA) with a simple 12-epoch cyclical learning rate schedule consistently improves object detection performance by about 1.0 AP across various models on the COCO benchmark, without additional inference costs.

Contribution

The study systematically applies SWA to object detection and segmentation, revealing a simple, effective policy that enhances accuracy without changing the detector architecture.

Findings

Achieves approximately 1.0 AP improvement across multiple detectors.

Effective application of SWA with 12 extra training epochs and cyclical learning rates.

No additional inference cost required.

Abstract

Do you want to improve 1.0 AP for your object detector without any inference cost and any change to your detector? Let us tell you such a recipe. It is surprisingly simple: train your detector for an extra 12 epochs using cyclical learning rates and then average these 12 checkpoints as your final detection model}. This potent recipe is inspired by Stochastic Weights Averaging (SWA), which is proposed in arXiv:1803.05407 for improving generalization in deep neural networks. We found it also very effective in object detection. In this technique report, we systematically investigate the effects of applying SWA to object detection as well as instance segmentation. Through extensive experiments, we discover the aforementioned workable policy of performing SWA in object detection, and we consistently achieve $\sim$1.0 AP improvement over various popular detectors on the challenging COCO benchmark, including Mask RCNN, Faster RCNN, RetinaNet, FCOS, YOLOv3 and VFNet. We hope this work will make more researchers in object detection know this technique and help them train better object detectors. Code is available at: https://github.com/hyz-xmaster/swa_object_detection .