Deep Transferring Quantization

We provide PyTorch implementation for "Deep Transferring Quantization" (ECCV2020).

Paper

Deep Transferring Quantization
Zheng Xie^*, Zhiquan Wen^*, Jing Liu^*, Zhiqiang Liu, Xixian Wu, and Mingkui Tan^*
European Conference on Computer Vision (ECCV), 2020

Dependencies

Python 3.6
PyTorch 1.1.0
dependencies in requirements.txt

Getting Started

Installation

Clone this repository:

 git clone https://github.com/xiezheng-cs/DTQ.git
 cd DTQ

Install PyTorch and other dependencies:
```
 pip install -r requirements.txt
```

Training

To quantize the pre-trained MobileNetV2 on Caltech 256-30 to 4-bit:

python main.py hocon_config/mobilenetv2_caltech-256-30.hocon

To quantize the pre-trained ResNet-50 on Caltech 256-30 to 4-bit:

python main.py hocon_config/resnet50_caltech-256-30.hocon

Experimental Results

Target Data Set	Model	W / A	DELTA-Q Top1 Acc(%)	DTQ(Ours) Top1 Acc(%)
Caltech 256-30	MobileNetV2	4 / 4	74.0±0.7	75.9±1.2
Caltech 256-30	ResNet-50	4 / 4	82.8±0.5	83.5±0.6

Citation

If this work is useful for your research, please cite our paper:

@InProceedings{xie2020deep,
title = {Deep Transferring Quantization},
author = {Zheng, Xie and Zhiquan, Wen and Jing, Liu and Zhiqiang, Liu and Xixian, Wu and Mingkui, Tan},
booktitle = {European Conference on Computer Vision (ECCV)},
year = {2020}
}

Acknowledgments

This work was partially supported by the Key-Area Research and Development Program of Guangdong Province 2019B010155002, National Natural Science Foundation of China (NSFC) 61836003 (key project), Program for Guangdong Introducing Innovative and Entrepreneurial Teams 2017ZT07X183, Fundamental Research Funds for the Central Universities D2191240.