arrow
返回

Just Recognizable Distortion for Machine Vision Oriented Image and Video Coding

delete2021-08-13
delete13
PRE
AI
Q
Qi Zhang *
王苫社 封面图
王苫社 (Shanshe Wang)
X
Xinfeng Zhang
马
马思伟 (Siwei Ma)
高
高雯 (Wen Gao)
DOI:10.1007/s11263-021-01505-4delete
delete原文链接
delete原文求助
delete分享
delete收藏
摘要

摘要

En 中文
Machine visual intelligence has exploded in recent years. Large-scale, high-quality image and video datasets significantly empower learning-based machine vision models, especially deep-learning models. However, images and videos are usually compressed before being analyzed in practical situations where transmission or storage is limited, leading to a noticeable performance loss of vision models. In this work, we broadly investigate the impact on the performance of machine vision from image and video coding. Based on the investigation, we propose Just Recognizable Distortion (JRD) to present the maximum distortion caused by data compression that will reduce the machine vision model performance to an unacceptable level. A large-scale JRD-annotated dataset containing over 340,000 images is built for various machine vision tasks, where the factors for different JRDs are studied. Furthermore, an ensemble-learning-based framework is established to predict the JRDs for diverse vision tasks under few- and non-reference conditions, which consists of multiple binary classifiers to improve the prediction accuracy. Experiments prove the effectiveness of the proposed JRD-guided image and video coding to significantly improve compression and machine vision performance. Applying predicted JRD is able to achieve remarkably better machine vision task accuracy and save a large number of bits.
Keyword:
Image and video coding
Machine vision
Deep learning
Just noticeable distortion
AI总结

AI总结

对已上传原文的论文进行重点信息的提取,主要内容包括:简要概述、研究摘要、背景介绍、关键亮点、图文解析、展望与总结。

期刊

International Journal of Computer Vision 封面图
International Journal of Computer Vision
IF:
9.3
论文数:
3.9K
被引数:
2.8W

机构

P
peking university
学者数:
11.9W
论文数: 8.7W
被引数: 146
C
chinese academy of sciences
学者数:
56.7W
论文数: 45.0W
被引数: 704
引用论文

引用论文

Studies on agalwood (Jinko). X. Structures of 2-(2-phenylethyl)chromone derivatives.
err1991-01-01
err0
errOAAI
errTenji KONISHI; Kiyoshi IWAGOE; Akira SUGIMOTO; Shiu KIYOSAWA; Yasuhiro FUJIWARA; Yasuo SHIMADA
err分享
err收藏
Identification of Enolase 1 and Thrombospondin-1 as serum biomarkers in HBV hepatic fibrosis by proteomics
err2013-07-11
err0
errOAAI
errBin Zhang; Zi Wang; Bin Deng; Xiaoqiong Wu; Jing Liu; Xueping Feng
err分享
err收藏
Global Warming Potential of Inhaled Anesthetics
err2010-07-01
err0
errOAAI
errSusan M. Ryan; Claus J. Nielsen
err分享
err收藏
Self-management programme for patients with long-term inflammatory bowel disease
err2011-04-01
err0
PREAI
errSharon Gethins; Tracey Duckett; Chris Shatford; Richard Robinson
err分享
err收藏
Synthesis of n-type semiconducting diamond film using diphosphorus pentaoxide as the doping source以五氧化二磷为掺杂源合成n型半导体金刚石膜
err1990-10-01
err0
PREAI
errKen Okano; Hideo Kiyota; Tatsuya Iwasaki; Yoshitaka Nakamura; Yukio Akiba; Tateki Kurosu; Masamori Iida; Terutaro Nakamura
err分享
err收藏
Quantitative phase microscopy with enhanced contrast and improved resolution through ultra‐oblique illumination (UO‐QPM)
err2019-06-30
err0
PREAI
errYing Ma; Siyue Guo; Yang Pan; Rong Fan; Zachary J. Smith; Stephen Lane; Kaiqin Chu
err分享
err收藏
学者 查看更多内容