Research Article
Promoting Animation Synthesis Through Dual-Channel Fusion
@INPROCEEDINGS{10.1007/978-3-031-55471-1_21, author={XiaoHong Qiu and ChaoChao Guo and Cong Xu}, title={Promoting Animation Synthesis Through Dual-Channel Fusion}, proceedings={Mobile Networks and Management. 13th EAI International Conference, MONAMI 2023, Yingtan, China, October 27-29, 2023, Proceedings}, proceedings_a={MONAMI}, year={2024}, month={3}, keywords={Animation Synthesis Channel Attention Dual-Channel Fusion}, doi={10.1007/978-3-031-55471-1_21} }- XiaoHong Qiu
ChaoChao Guo
Cong Xu
Year: 2024
Promoting Animation Synthesis Through Dual-Channel Fusion
MONAMI
Springer
DOI: 10.1007/978-3-031-55471-1_21
Abstract
Although animation synthesis technology is widely applied, it also imposes higher demands on the precision of the synthesized animation. This paper employs a more lightweight channel attention module for image feature extraction. Compared to previous channel attention module, this approach utilizes fewer parameters, thereby assisting the network in achieving improved precision. Additionally, it replaces the sigmoid function with the more suitable output function tanh for image generation. Three evaluation metrics show improvements: a 1.3% increase in L1, an 18.9% increase in AED, and a 2.6% increase in AKD. To facilitate better image generation by the generator, improvements are made to the discriminator. Spectral normalization and instance normalization are combined to form a multi-normalization module for normalization during the image down sampling process. Additionally, an adaptive Dual-Channel Fusion output module is employed for the discriminator output, aiding in the rapid convergence of the network. The quality metrics of the generated images demonstrate improvements, with a 4.3% increase in L1, a 23.8% increase in AED, and a 5.5% increase in AKD.
