Skip to content

Latest commit

 

History

History
 
 

solov2

SOLOv2

SOLOv2: Dynamic and Fast Instance Segmentation

Abstract

In this work, we aim at building a simple, direct, and fast instance segmentation framework with strong performance. We follow the principle of the SOLO method of Wang et al. "SOLO: segmenting objects by locations". Importantly, we take one step further by dynamically learning the mask head of the object segmenter such that the mask head is conditioned on the location. Specifically, the mask branch is decoupled into a mask kernel branch and mask feature branch, which are responsible for learning the convolution kernel and the convolved features respectively. Moreover, we propose Matrix NMS (non maximum suppression) to significantly reduce the inference time overhead due to NMS of masks. Our Matrix NMS performs NMS with parallel matrix operations in one shot, and yields better results. We demonstrate a simple direct instance segmentation system, outperforming a few state-of-the-art methods in both speed and accuracy. A light-weight version of SOLOv2 executes at 31.3 FPS and yields 37.1% AP. Moreover, our state-of-the-art results in object detection (from our mask byproduct) and panoptic segmentation show the potential to serve as a new strong baseline for many instance-level recognition tasks besides instance segmentation.

Results and Models

SOLOv2

Backbone Style MS train Lr schd Mem (GB) mask AP Config Download
R-50 pytorch N 1x 5.1 34.8 config model | log
R-50 pytorch Y 3x 5.1 37.5 config model | log
R-101 pytorch Y 3x 6.9 39.1 config model | log
R-101(DCN) pytorch Y 3x 7.1 41.2 config model | log
X-101(DCN) pytorch Y 3x 11.3 42.4 config model | log

Light SOLOv2

Backbone Style MS train Lr schd Mem (GB) mask AP Config Download
R-18 pytorch Y 3x 9.1 29.7 config model | log
R-34 pytorch Y 3x 9.3 31.9 config model | log
R-50 pytorch Y 3x 9.9 33.7 config model | log

Citation

@article{wang2020solov2,
  title={SOLOv2: Dynamic and Fast Instance Segmentation},
  author={Wang, Xinlong and Zhang, Rufeng and  Kong, Tao and Li, Lei and Shen, Chunhua},
  journal={Proc. Advances in Neural Information Processing Systems (NeurIPS)},
  year={2020}
}