NVS-SQA: Exploring Self-Supervised Quality Representation Learning for Neurally Synthesized Scenes Without References

过度拟合 计算机科学 人工智能 边距(机器学习) 质量(理念) 机器学习 代表(政治) 启发式 过程(计算) 感知 质量得分 质量评定 模式识别(心理学) 高斯过程 特征学习 众包 卷积神经网络 启发式 任务分析 深度学习 计算机视觉 特征提取 稀疏逼近 训练集
作者
Qiang Qu,Yiran Shen,Xiaoming Chen,Yuk Ying Chung,Weidong Cai,Tongliang Liu
出处
期刊:IEEE Transactions on Pattern Analysis and Machine Intelligence [IEEE Computer Society]
卷期号:48 (3): 2265-2281 被引量:10
标识
DOI:10.1109/tpami.2025.3626550
摘要

Neural View Synthesis (NVS), such as NeRF and 3D Gaussian Splatting, effectively creates photorealistic scenes from sparse viewpoints, typically evaluated by quality assessment methods like PSNR, SSIM, and LPIPS. However, these full-reference methods, which compare synthesized views to reference views, may not fully capture the perceptual quality of neurally synthesized scenes (NSS), particularly due to the limited availability of dense reference views. Furthermore, the challenges in acquiring human perceptual labels hinder the creation of extensive labeled datasets, risking model overfitting and reduced generalizability. To address these issues, we propose NVS-SQA, a NSS quality assessment method to learn no-reference quality representations through self-supervision without reliance on human labels. Traditional self-supervised learning predominantly relies on the "same instance, similar representation" assumption and extensive datasets. However, given that these conditions do not apply in NSS quality assessment, we employ heuristic cues and quality scores as learning objectives, along with a specialized contrastive pair preparation process to improve the effectiveness and efficiency of learning. The results show that NVS-SQA outperforms 17 no-reference methods by a large margin (i.e., on average 109.5% in SRCC, 98.6% in PLCC, and 91.5% in KRCC over the second best) and even exceeds 16 full-reference methods across all evaluation metrics (i.e., 22.9% in SRCC, 19.1% in PLCC, and 18.6% in KRCC over the second best).

科研通智能强力驱动
Strongly Powered by AbleSci AI
科研通是完全免费的文献互助平台,具备全网最快的应助速度,最高的求助完成率。 对每一个文献求助,科研通都将尽心尽力,给求助人一个满意的交代。
实时播报
顾矜的应助被小y同学采纳,获得10
刚刚
刚刚
minmin完成签到 ,获得积分10
1秒前
在水一方的应助被Eana采纳,获得10
2秒前
2秒前
2秒前
妮妮完成签到,获得积分10
2秒前
3秒前
大个的应助被科研通管家采纳,获得10
3秒前
Cr的应助被科研通管家采纳,获得10
3秒前
科研通AI2S的应助被临平吴彦祖采纳,获得10
3秒前
CipherSage的应助被科研通管家采纳,获得10
3秒前
sendou的应助被科研通管家采纳,获得30
3秒前
Jasper的应助被科研通管家采纳,获得10
3秒前
4秒前
yn107发布了新的文献求助10
4秒前
顾矜的应助被科研通管家采纳,获得10
4秒前
科研菜菜的应助被科研通管家采纳,获得30
4秒前
Eric完成签到,获得积分10
4秒前
4秒前
wanci的应助被科研通管家采纳,获得10
4秒前
英俊的铭的应助被科研通管家采纳,获得10
4秒前
上官若男的应助被科研通管家采纳,获得10
4秒前
脑洞疼的应助被科研通管家采纳,获得10
5秒前
Akim的应助被科研通管家采纳,获得10
5秒前
所所的应助被科研通管家采纳,获得10
5秒前
5秒前
Jasper的应助被科研通管家采纳,获得10
5秒前
Cr的应助被科研通管家采纳,获得10
5秒前
思源的应助被科研通管家采纳,获得10
5秒前
Orange的应助被xyt采纳,获得10
5秒前
Cr的应助被科研通管家采纳,获得10
5秒前
无奈醉柳完成签到,获得积分10
6秒前
qian完成签到,获得积分10
6秒前
7秒前
小方发布了新的文献求助10
7秒前
水弘的应助被暮晓见采纳,获得10
8秒前
9秒前
wiliiam完成签到,获得积分10
9秒前
10秒前
高分求助中
(应助此贴封号)【重要!!请各用户(尤其是新用户)详细阅读】【科研通的精品贴汇总】 10000
拉伸高分子熔体中缠结行为的模拟和小角中子散射研究 1700
The World Atlas of Wine 9th edition 1000
奈特基础组织学:配套组织病理学(第4版)(Netter's Essential Histology With Correlated Histopathology) 800
八旗制度の研究 600
Improving the wear resistance of vehicle gear oil by oil soluble LaF3 nanoparticles and typical additives of gear oil 500
Boundaries and Bodhisattvas 500
热门求助领域 (近24小时)
化学 材料科学 医学 生物 纳米技术 计算机科学 工程类 有机化学 化学工程 生物化学 物理 复合材料 内科学 催化作用 光电子学 人工智能 细胞生物学 无机化学 物理化学 数学
热门帖子
关注 科研通微信公众号,转发送积分 8183508
求助须知:如何正确求助?哪些是违规求助? 9572862
关于积分的说明 21827639
捐赠科研通 7780681
什么是DOI,文献DOI怎么找? 3434016
关于科研通互助平台的介绍 2558189
邀请新用户注册赠送积分活动 2480405