Sparse R-CNN: End-to-End Object Detection with Learnable Proposals

TL;DR

Sparse R-CNN introduces a novel end-to-end object detection method that uses a fixed set of learnable proposals, eliminating the need for dense candidate generation and post-processing, achieving competitive accuracy and speed.

Contribution

It proposes a fully sparse, learnable proposal-based object detection framework that simplifies design and improves efficiency over traditional dense methods.

Findings

Achieves 45.0 AP on COCO with ResNet-50 FPN.

Operates at 22 fps, comparable to established detectors.

Eliminates non-maximum suppression in detection pipeline.

Abstract

We present Sparse R-CNN, a purely sparse method for object detection in images. Existing works on object detection heavily rely on dense object candidates, such as $k$ anchor boxes pre-defined on all grids of image feature map of size $H\times W$. In our method, however, a fixed sparse set of learned object proposals, total length of $N$, are provided to object recognition head to perform classification and location. By eliminating $HWk$ (up to hundreds of thousands) hand-designed object candidates to $N$ (e.g. 100) learnable proposals, Sparse R-CNN completely avoids all efforts related to object candidates design and many-to-one label assignment. More importantly, final predictions are directly output without non-maximum suppression post-procedure. Sparse R-CNN demonstrates accuracy, run-time and training convergence performance on par with the well-established detector baselines on the challenging COCO dataset, e.g., achieving 45.0 AP in standard $3\times$ training schedule and running at 22 fps using ResNet-50 FPN model. We hope our work could inspire re-thinking the convention of dense prior in object detectors. The code is available at: https://github.com/PeizeSun/SparseR-CNN.