Arbitrary Video Style Transfer via Multi-Channel Correlation

Authors: Yingying Deng, Fan Tang, Weiming Dong, Haibin Huang, Chongyang Ma, Changsheng Xu1210-1217

AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Qualitative and quantitative evaluations demonstrate that MCCNet performs well in arbitrary video and image style transfer tasks.
Researcher Affiliation Collaboration 1 School of Artificial Intelligence, University of Chinese Academy of Sciences, 2 NLPR, Institute of Automation, Chinese Academy of Sciences, 3 School of Artificial Intelligence, Jilin University, 4 CASIA-LLvision Joint Lab, 5 Kuaishou Technology
Pseudocode No The paper describes the multi-channel correlation and network structure but does not include any pseudocode or clearly labeled algorithm blocks.
Open Source Code Yes Code is available at https://github.com/diyiiyiii/MCCNet.
Open Datasets Yes We use MS-COCO (Lin et al. 2014) and Wiki Art (Phillips and Mackintosh 2011) as the content and style image datasets for network training.
Dataset Splits No The paper mentions using MS-COCO and Wiki Art for training but does not specify clear training/validation/test splits (e.g., percentages or exact counts) for reproducibility.
Hardware Specification Yes We measure our inference time for the generation of an output image and compare the result with those of SOTA methods using 16G Titan X GPU.
Software Dependencies No The paper does not provide specific version numbers for software dependencies (e.g., programming languages or libraries).
Experiment Setup Yes The weights λcontent, λstyle, λid, and λillum are set to 4, 15, 70, and 3, 000 to eliminate the impact of magnitude differences. ... The training batch size is 8, and the whole model is trained through 160, 000 steps.