Few-shot Neural Human Performance Rendering from Sparse RGBD Videos
Authors: Anqi Pang, Xin Chen, Haimin Luo, Minye Wu, Jingyi Yu, Lan Xu
IJCAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments demonstrate the effectiveness of our approach to generate high-quality free view-point results for challenging human performances under the sparse setting. |
| Researcher Affiliation | Academia | Anqi Pang1,2,3 , Xin Chen1,2,3 , Haimin Luo1,2,3 , Minye Wu1,2,3 , Jingyi Yu1 , Lan Xu1 1 Shanghai Engineering Research Center of Intelligent Vision and Imaging, School of Information Science and Technology, Shanghai Tech University 2 Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences 3 University of Chinese Academy of Sciences |
| Pseudocode | No | The paper does not contain any clearly labeled pseudocode or algorithm blocks. It provides architectural diagrams but not code-like procedures. |
| Open Source Code | No | The paper does not provide an explicit statement or link for the release of its source code. |
| Open Datasets | No | The paper states: 'Our approach takes only 6 RGBD streams from stereo cameras or Azure Kinect sensors surrounding the performer as input'. It does not refer to a publicly available dataset with concrete access information (link, citation, or repository). |
| Dataset Splits | No | The paper mentions training data and key-frame selection but does not provide specific details about validation dataset splits (e.g., percentages, sample counts, or methodology for splitting). |
| Hardware Specification | Yes | We run our experiments on a PC with 2.2 GHz Intel Xeon 4210 CPU 64GB RAM, and Nvidia TITAN RTX GPU. |
| Software Dependencies | No | The paper mentions software components like U-Net, PointNet++, Deep Lab v3, and OpenPose, but it does not specify version numbers for any of these to ensure reproducibility. |
| Experiment Setup | Yes | We set them to 0.3, 5, and 0.7 respectively in our experiments. We use the Adam optimizer with a learning rate of 0.0002 to optimize the network parameters. The batch size is 4, and the number of one batch sample is 40. We also augment our training data with random translation, random scaling, and random rotation. |