WaveFormer: Wavelet Transformer for Noise-Robust Video Inpainting

修补 变压器 小波 计算机科学 人工智能 计算机视觉 工程类 图像(数学) 电气工程 电压
作者
Zhiliang Wu,Changchang Sun,Hanyu Xuan,Gaowen Liu,Yan Yan
出处
期刊:Proceedings of the ... AAAI Conference on Artificial Intelligence [Association for the Advancement of Artificial Intelligence (AAAI)]
卷期号:38 (6): 6180-6188 被引量:4
标识
DOI:10.1609/aaai.v38i6.28435
摘要

Video inpainting aims to fill in the missing regions of the video frames with plausible content. Benefiting from the outstanding long-range modeling capacity, the transformer-based models have achieved unprecedented performance regarding inpainting quality. Essentially, coherent contents from all the frames along both spatial and temporal dimensions are concerned by a patch-wise attention module, and then the missing contents are generated based on the attention-weighted summation. In this way, attention retrieval accuracy has become the main bottleneck to improve the video inpainting performance, where the factors affecting attention calculation should be explored to maximize the advantages of transformer. Towards this end, in this paper, we theoretically certificate that noise is the culprit that entangles the process of attention calculation. Meanwhile, we propose a novel wavelet transformer network with noise robustness for video inpainting, named WaveFormer. Unlike existing transformer-based methods that utilize the whole embeddings to calculate the attention, our WaveFormer first separates the noise existing in the embedding into high-frequency components by introducing the Discrete Wavelet Transform (DWT), and then adopts clean low-frequency components to calculate the attention. In this way, the impact of noise on attention computation can be greatly mitigated and the missing content regarding different frequencies can be generated by sharing the calculated attention. Extensive experiments validate the superior performance of our method over state-of-the-art baselines both qualitatively and quantitatively.

科研通智能强力驱动
Strongly Powered by AbleSci AI
更新
大幅提高文件上传限制,最高150M (2024-4-1)

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
lee发布了新的文献求助30
3秒前
6秒前
Zzzooo完成签到,获得积分20
7秒前
zwd完成签到,获得积分10
7秒前
JoaquinH完成签到,获得积分10
7秒前
Smile2044完成签到,获得积分20
8秒前
9秒前
LYH发布了新的文献求助10
9秒前
傲娇大船完成签到,获得积分10
10秒前
Tuotuo完成签到 ,获得积分10
10秒前
852应助莀莀采纳,获得10
11秒前
evelynjm发布了新的文献求助10
13秒前
科研通AI2S应助dan1029采纳,获得10
13秒前
嗯哼应助dan1029采纳,获得10
13秒前
Akim应助大力日记本采纳,获得10
13秒前
科研通AI2S应助dan1029采纳,获得10
13秒前
monere应助相信明天会更好采纳,获得10
15秒前
大个应助耳东陈采纳,获得30
15秒前
我是老大应助哈哈采纳,获得10
16秒前
Dollar完成签到,获得积分10
17秒前
萧水白应助aaaaa采纳,获得10
17秒前
20秒前
20秒前
科研吗喽发布了新的文献求助10
21秒前
月满西楼完成签到,获得积分10
21秒前
Parkryeol完成签到,获得积分10
22秒前
善学以致用应助不二仙采纳,获得10
23秒前
monere应助今天开心吗采纳,获得10
23秒前
Dollar发布了新的文献求助10
24秒前
李健应助cleva采纳,获得10
27秒前
evelynjm完成签到,获得积分10
28秒前
29秒前
在路上发布了新的文献求助10
30秒前
30秒前
和谐的曼云完成签到,获得积分10
33秒前
刘yuer完成签到,获得积分10
34秒前
smy完成签到,获得积分10
34秒前
传奇3应助木棉采纳,获得10
35秒前
wang发布了新的文献求助10
35秒前
lin发布了新的文献求助50
35秒前
高分求助中
The late Devonian Standard Conodont Zonation 2000
歯科矯正学 第7版(或第5版) 1004
Nickel superalloy market size, share, growth, trends, and forecast 2023-2030 1000
Semiconductor Process Reliability in Practice 1000
Smart but Scattered: The Revolutionary Executive Skills Approach to Helping Kids Reach Their Potential (第二版) 1000
Security Awareness: Applying Practical Cybersecurity in Your World 6th Edition 800
PraxisRatgeber: Mantiden: Faszinierende Lauerjäger 700
热门求助领域 (近24小时)
化学 医学 生物 材料科学 工程类 有机化学 生物化学 物理 内科学 纳米技术 计算机科学 化学工程 复合材料 基因 遗传学 催化作用 物理化学 免疫学 量子力学 细胞生物学
热门帖子
关注 科研通微信公众号,转发送积分 3240186
求助须知:如何正确求助?哪些是违规求助? 2885221
关于积分的说明 8237360
捐赠科研通 2553498
什么是DOI,文献DOI怎么找? 1381664
科研通“疑难数据库(出版商)”最低求助积分说明 649317
邀请新用户注册赠送积分活动 625009