TY - GEN
T1 - Efficient Semantic-Guidance High-Resolution Video Matting
AU - Yu, Yue
AU - Li, Ding
AU - Yang, Yulin
N1 - Publisher Copyright:
© 2024, The Author(s), under exclusive license to Springer Nature Switzerland AG.
PY - 2024
Y1 - 2024
N2 - Video matting has made significant progress in trimap-based field. However, researchers are increasingly interested in auxiliary-free matting because it is more useful in real-world applications. We propose a new efficient semantic-guidance high-resolution video matting network for human body. We apply the convolutional network as the backbone while also employing the transformer in the encoder, which is used to utilize semantic features, while ensuring that the network is not overly bloated. In addition, a channel-wise attention mechanism is introduced in the decoder to improve the representation of semantic feature. In comparison to the current state-of-the-art methods, the method proposed in this paper achieves better results while maintaining the speed and efficiency of prediction. We can complete the real-time auxiliary-free matting for high-resolution video (4K or HD).
AB - Video matting has made significant progress in trimap-based field. However, researchers are increasingly interested in auxiliary-free matting because it is more useful in real-world applications. We propose a new efficient semantic-guidance high-resolution video matting network for human body. We apply the convolutional network as the backbone while also employing the transformer in the encoder, which is used to utilize semantic features, while ensuring that the network is not overly bloated. In addition, a channel-wise attention mechanism is introduced in the decoder to improve the representation of semantic feature. In comparison to the current state-of-the-art methods, the method proposed in this paper achieves better results while maintaining the speed and efficiency of prediction. We can complete the real-time auxiliary-free matting for high-resolution video (4K or HD).
KW - Attention Mechanism
KW - Auxiliary-Free Video Matting Network
KW - Transformer
KW - Video Matting
UR - http://www.scopus.com/inward/record.url?scp=85184287174&partnerID=8YFLogxK
U2 - 10.1007/978-3-031-50069-5_13
DO - 10.1007/978-3-031-50069-5_13
M3 - Conference contribution
AN - SCOPUS:85184287174
SN - 9783031500688
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 143
EP - 154
BT - Advances in Computer Graphics - 40th Computer Graphics International Conference, CGI 2023, Proceedings
A2 - Sheng, Bin
A2 - Bi, Lei
A2 - Kim, Jinman
A2 - Magnenat-Thalmann, Nadia
A2 - Thalmann, Daniel
PB - Springer Science and Business Media Deutschland GmbH
T2 - 40th Computer Graphics International Conference, CGI 2023
Y2 - 28 August 2023 through 1 September 2023
ER -