Skip to content
Branch: master
Find file History
ppwwyyxx and facebook-github-bot Use explicit versions in builtin configs (#363)
Currently loading any official config will contain a warning of missing VERSION.
Pull Request resolved: fairinternal/detectron2#363

Reviewed By: rbgirshick

Differential Revision: D19260690

Pulled By: ppwwyyxx

fbshipit-source-id: 2b17d6ed34aa3118c8bb89d40952e9d00054758d
Latest commit 54d8e79 Jan 8, 2020
Type Name Latest commit message Commit time
Failed to load latest commit information.
configs Use explicit versions in builtin configs (#363) Jan 8, 2020
tensormask update docs Dec 14, 2019
tests support empty inputs for retinanet & add tests Nov 30, 2019 Add a pre-trained model Dec 2, 2019 TMask: add swap_align2nat op. (#266) Nov 26, 2019 TMask: Main (#268) Nov 26, 2019

TensorMask in Detectron2

A Foundation for Dense Object Segmentation

Xinlei Chen, Ross Girshick, Kaiming He, Piotr Dollár

[arXiv] [BibTeX]

In this repository, we release code for TensorMask in Detectron2. TensorMask is a dense sliding-window instance segmentation framework that, for the first time, achieves results close to the well-developed Mask R-CNN framework -- both qualitatively and quantitatively. It establishes a conceptually complementary direction for object instance segmentation research.


First install Detectron 2 following Then compile the TensorMask-specific op (swap_align2nat):

cd /path/to/detectron2/projects/TensorMask
python build develop


To train a model, run:

python /path/to/detectron2/projects/TensorMask/ --config-file <config.yaml>

For example, to launch TensorMask BiPyramid training (1x schedule) with ResNet-50 backbone on 8 GPUs, one should execute:

python /path/to/detectron2/projects/TensorMask/ --config-file configs/tensormask_R_50_FPN_1x.yaml --num-gpus 8


Model evaluation can be done similarly (6x schedule with scale augmentation):

python /path/to/detectron2/projects/TensorMask/ --config-file configs/tensormask_R_50_FPN_6x.yaml --eval-only MODEL.WEIGHTS /path/to/model_checkpoint

Pretrained Models

Backbone lr sched AP box AP mask download
R50 1x 37.6 32.4 model |  metrics
R50 6x 41.4 35.8 model |  metrics

Citing TensorMask

If you use TensorMask, please use the following BibTeX entry.

  title={Tensormask: A Foundation for Dense Object Segmentation},
  author={Chen, Xinlei and Girshick, Ross and He, Kaiming and Doll{\'a}r, Piotr},
  journal={The International Conference on Computer Vision (ICCV)},
You can’t perform that action at this time.