Cross Domain Model Compression by Structurally Weight Sharing

Shangqian Gao, Cheng Deng, Heng Huang; The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2019, pp. 8973-8982

Abstract


Regular model compression methods focus on RGB input. While cross domain tasks demand more DNN models, each domain often needs its own model. Consequently, for such tasks, the storage cost, memory footprint and computation cost increase dramatically compared to single RGB input. Moreover, the distinct appearance and special structure in cross domain tasks make it difficult to directly apply regular compression methods on it. In this paper, thus, we propose a new robust cross domain model compression method. Specifically, the proposed method compress cross domain models by structurally weight sharing, which is achieved by regularizing the models with graph embedding at training time. Due to the channel wise weights sharing, the proposed method can reduce computation cost without specially designed algorithm. In the experiments, the proposed method achieves state of the art results on two diverse tasks: action recognition and RGB-D scene recognition.

Related Material


[pdf] [supp]
[bibtex]
@InProceedings{Gao_2019_CVPR,
author = {Gao, Shangqian and Deng, Cheng and Huang, Heng},
title = {Cross Domain Model Compression by Structurally Weight Sharing},
booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2019}
}