Thinking in Granularity: Dynamic Quantization for Image Super-Resolution by Intriguing Multi-Granularity Clues

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wang, Mingshen, Zhang, Zhao, Li, Feng, Xu, Ke, Miao, Kang, Wang, Meng
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912164873240576
author Wang, Mingshen
Zhang, Zhao
Li, Feng
Xu, Ke
Miao, Kang
Wang, Meng
author_facet Wang, Mingshen
Zhang, Zhao
Li, Feng
Xu, Ke
Miao, Kang
Wang, Meng
contents Dynamic quantization has attracted rising attention in image super-resolution (SR) as it expands the potential of heavy SR models onto mobile devices while preserving competitive performance. Existing methods explore layer-to-bit configuration upon varying local regions, adaptively allocating the bit to each layer and patch. Despite the benefits, they still fall short in the trade-off of SR accuracy and quantization efficiency. Apart from this, adapting the quantization level for each layer individually can disturb the original inter-layer relationships, thus diminishing the representation capability of quantized models. In this work, we propose Granular-DQ, which capitalizes on the intrinsic characteristics of images while dispensing with the previous consideration for layer sensitivity in quantization. Granular-DQ conducts a multi-granularity analysis of local patches with further exploration of their information densities, achieving a distinctive patch-wise and layer-invariant dynamic quantization paradigm. Specifically, Granular-DQ initiates by developing a granularity-bit controller (GBC) to apprehend the coarse-to-fine granular representations of different patches, matching their proportional contribution to the entire image to determine the proper bit-width allocation. On this premise, we investigate the relation between bit-width and information density, devising an entropy-to-bit (E2B) mechanism that enables further fine-grained dynamic bit adaption of high-bit patches. Extensive experiments validate the superiority and generalization ability of Granular-DQ over recent state-of-the-art methods on various SR models. Code and supplementary statement can be found at \url{https://github.com/MmmingS/Granular-DQ.git}.
format Preprint
id arxiv_https___arxiv_org_abs_2409_14330
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Thinking in Granularity: Dynamic Quantization for Image Super-Resolution by Intriguing Multi-Granularity Clues
Wang, Mingshen
Zhang, Zhao
Li, Feng
Xu, Ke
Miao, Kang
Wang, Meng
Image and Video Processing
Computer Vision and Pattern Recognition
Dynamic quantization has attracted rising attention in image super-resolution (SR) as it expands the potential of heavy SR models onto mobile devices while preserving competitive performance. Existing methods explore layer-to-bit configuration upon varying local regions, adaptively allocating the bit to each layer and patch. Despite the benefits, they still fall short in the trade-off of SR accuracy and quantization efficiency. Apart from this, adapting the quantization level for each layer individually can disturb the original inter-layer relationships, thus diminishing the representation capability of quantized models. In this work, we propose Granular-DQ, which capitalizes on the intrinsic characteristics of images while dispensing with the previous consideration for layer sensitivity in quantization. Granular-DQ conducts a multi-granularity analysis of local patches with further exploration of their information densities, achieving a distinctive patch-wise and layer-invariant dynamic quantization paradigm. Specifically, Granular-DQ initiates by developing a granularity-bit controller (GBC) to apprehend the coarse-to-fine granular representations of different patches, matching their proportional contribution to the entire image to determine the proper bit-width allocation. On this premise, we investigate the relation between bit-width and information density, devising an entropy-to-bit (E2B) mechanism that enables further fine-grained dynamic bit adaption of high-bit patches. Extensive experiments validate the superiority and generalization ability of Granular-DQ over recent state-of-the-art methods on various SR models. Code and supplementary statement can be found at \url{https://github.com/MmmingS/Granular-DQ.git}.
title Thinking in Granularity: Dynamic Quantization for Image Super-Resolution by Intriguing Multi-Granularity Clues
topic Image and Video Processing
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2409.14330