ArtechStark / GCNet

GCNet: Non-local Networks Meet Squeeze-Excitation Networks and Beyond

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

GCNet for Object Detection

By Yue Cao, Jiarui Xu, Stephen Lin, Fangyun Wei, Han Hu.

This repo is a official implementation of "GCNet: Non-local Networks Meet Squeeze-Excitation Networks and Beyond" on COCO object detection based on open-mmlab's mmdetection. Many thanks to mmdetection for their simple and clean framework.


GCNet is initially described in arxiv. Via absorbing advantages of Non-Local Networks (NLNet) and Squeeze-Excitation Networks (SENet), GCNet provides a simple, fast and effective approach for global context modeling, which generally outperforms both NLNet and SENet on major benchmarks for various recognition tasks.

Citing GCNet

  title={GCNet: Non-local Networks Meet Squeeze-Excitation Networks and Beyond},
  author={Cao, Yue and Xu, Jiarui and Lin, Stephen and Wei, Fangyun and Hu, Han},
  journal={arXiv preprint arXiv:1904.11492},

Main Results

Backbone Model Context Lr schd box AP mask AP Download
R-50-FPN Mask - 1x 37.2 33.8 model
GC(c3-c5, r16) 1x 39.4 35.7 model
GC(c3-c5, r4) 1x 39.9 36.2 model
R-101-FPN Mask - 1x 39.8 36.0 model
GC(c3-c5, r16) 1x 41.1 37.4 model
GC(c3-c5, r4) 1x 41.7 37.6 model
X-101-FPN Mask - 1x 41.2 37.3 model
GC(c3-c5, r16) 1x 42.4 38.0 model
GC(c3-c5, r4) 1x 42.9 38.5 model
X-101-FPN Cascade Mask - 1x 44.7 38.3 model
GC(c3-c5, r16) 1x 45.9 39.3 model
GC(c3-c5, r4) 1x 46.5 39.7 model
X-101-FPN DCN Cascade Mask - 1x 47.1 40.4 model
GC(c3-c5, r16) 1x 47.9 40.9 model
GC(c3-c5, r4) 1x 47.9 40.8 model


  • GC denotes Global Context (GC) block is inserted after 1x1 conv of backbone.
  • DCN denotes replace 3x3 conv with 3x3 Deformable Convolution in c3-c5 stages of backbone.
  • r4 and r16 denote ratio 4 and ratio 16 in GC block respectively.


  • Linux(tested on Ubuntu 16.04)
  • Python 3.6+
  • PyTorch 1.0.0
  • Cython
  • apex (Sync BN)


a. Install PyTorch 1.0 and torchvision following the official instructions.

b. Install latest apex with CUDA and C++ extensions following this instructions. The Sync BN implemented by apex is required.

c. Clone the GCNet repository.

 git clone 

d. Compile cuda extensions.

cd GCNet
pip install cython  # or "conda install cython" if you prefer conda
./  # or "PYTHON=python3 ./" if you use system python3 without virtual environments

e. Install GCNet version mmdetection (other dependencies will be installed automatically).

python(3) install  # add --user if you want to install it locally
# or "pip install ."

Note: You need to run the last step each time you pull updates from github. Or you can run python(3) develop or pip install -e . to install mmdetection if you want to make modifications to it frequently.

Please refer to mmdetection install instruction for more details.



  • 8 NVIDIA Tesla V100 GPUs
  • Intel Xeon 4114 CPU @ 2.20GHz

Software environment

  • Python 3.6.7
  • PyTorch 1.0.0
  • CUDA 9.0
  • CUDNN 7.0
  • NCCL 2.3.5



As in original mmdetection, distributed training is recommended for either single machine or multiple machines.

./tools/ <CONFIG_FILE> <GPU_NUM> [optional arguments]

Supported arguments are:

  • --validate: perform evaluation every k (default=1) epochs during the training.
  • --work_dir <WORK_DIR>: if specified, the path in config file will be replaced.


To evaluate trained models, output file is required.

python tools/ <CONFIG_FILE> <MODEL_PATH> [optional arguments]

Supported arguments are:

  • --gpus: number of GPU used for evaluation
  • --out: output file name, usually ends wiht .pkl
  • --eval: type of evaluation need, for mask-rcnn, bbox segm would evaluate both bounding box and mask AP.
ezoic increase your site revenue


GCNet: Non-local Networks Meet Squeeze-Excitation Networks and Beyond

License:Apache License 2.0


Language:Python 83.1%Language:Cuda 11.2%Language:C++ 5.6%Language:Shell 0.1%