iVMCL-Release

It is mainly built on two great python and PyTorch libraries: MMCV (commit fe83261, 10/01/2020) and MMDetection (commit 5b18b94, 10/01/2020). We will synchronize and update the two repos with the remote master repos when significant updates arise.

It also uses codes from the great PyToch Image Models Github repo.

It includes official PyTorch implementations of

AOGNets (CVPR2019) for image classification in ImageNet-1000 and object detection and semantic segmentation in MS-COCO. See the previous implementation at AOGNet-V2.
Attentive Normalization(ECCV2020) for image classification in ImageNet-1000 and object detection and semantic segmentation in MS-COCO. See the previous implementation at AttentiveNorm_Detection based on MMDetection v1.x.
Interpretable R-CNN (ICCV2019) for object detection in PASCAL VOC07 and MS-COCO. To be released a.s.a.p. Please stay tuned.

Model Zoo

ImageNet Classification
MS-COCO Detection & Segmentation In progress.

Training from Scratch

Installation

It follows the installation instructions in MMCV and MMDetection, which are summarized as follows.

Requirements

Linux or macOS (Windows is not currently officially supported)
Python 3.6+ (Anacoda is recommended)
PyTorch 1.3+
CUDA 9.2+ (If you build PyTorch from source, CUDA 9.0 is also compatible)
GCC 5+
NVIDIA Apex, which has been integrated in PyTorch 1.6+.

Install iVMCL-Release

a. Create a conda virtual environment and activate it.

conda create -n ivmcl-release python=3.7 -y
conda activate ivmcl-release

b. Install PyTorch and torchvision following the official instructions, e.g.,

conda install pytorch torchvision -c pytorch

Note: Make sure that your compilation CUDA version and runtime CUDA version match. You can check the supported CUDA version for precompiled packages on the PyTorch website.

E.g.1 If you have CUDA 10.1 installed under /usr/local/cuda and would like to install PyTorch 1.5, you need to install the prebuilt PyTorch with CUDA 10.1.

conda install pytorch cudatoolkit=10.1 torchvision -c pytorch

E.g. 2 If you have CUDA 9.2 installed under /usr/local/cuda and would like to install PyTorch 1.3.1., you need to install the prebuilt PyTorch with CUDA 9.2.

conda install pytorch=1.3.1 cudatoolkit=9.2 torchvision=0.4.2 -c pytorch

If you build PyTorch from source instead of installing the prebuilt pacakge, you can use more CUDA versions such as 9.0.

c. Clone the iVMCL-Release repository.

git clone https://github.com/iVMCL/iVMCL-Release.git

d. Compile mmcv

cd iVMCL-Release/mmcv
MMCV_WITH_OPS=1 pip install -e .  # package mmcv-full will be installed after this step
cd ..

e. Compile mmdetection

cd mmdetection
pip install -r requirements/build.txt
pip install -r requirements/ivmcl.txt
pip install -v -e .  # or "python setup.py develop"

If you build mmdetection on macOS, replace the last command with

CC=clang CXX=clang++ CFLAGS='-stdlib=libc++' pip install -e .

f. Compile NVIDIA apex (for image classification)

cd YOUR_PATH_TO/apex
pip install -v --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" ./

g. Update Pillow-SIMD for faster image IO

cd mmdetection/scripts_ivmcl
chmod +x ./update_pillow.sh
./update_pillow.sh

Citations

Please consider citing the following papers in your publications if they help your research.

@inproceedings{li2019aognets,
  title={AOGNets: Compositional Grammatical Architectures for Deep Learning},
  author={Li, Xilai and Song, Xi and Wu, Tianfu},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition {(CVPR)}},
  pages={6220--6230},
  year={2019}
}

@inproceedings{li2020attentive,
  title={Attentive Normalization},
  author={Li, Xilai and Sun, Wei and Wu, Tianfu},
  booktitle={Proceedings of the European Conference on Computer Vision {(ECCV)}},
  year={2020}
}

@inproceedings{wu2019towards,
  title={Towards Interpretable Object Detection by Unfolding Latent Structures},
  author={Wu, Tianfu and Song, Xi},
  booktitle={Proceedings of the IEEE International Conference on Computer Vision {(ICCV)}},
  pages={6033--6043},
  year={2019}
}

Contact

Please feel free to report issues and any related problems to Tianfu Wu (twu19 at ncsu dot edu).

License

AOGNets related codes are under RESEARCH ONLY LICENSE.

tfwu / iVMCL-Release