SparseGNV: Generating Novel Views of Indoor Scenes with Sparse RGB-D Images

Authors: Weihao Cheng, Yan-Pei Cao, Ying Shan

AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We evaluate Sparse GNV on real-world indoor scenes and demonstrate that it outperforms state-of-the-art methods based on either neural radiance fields or conditional image generation. As the results presented in Table 1, our method outperforms all the baselines on PSNR, SSIM, and LPIPS.
Researcher Affiliation Industry Weihao Cheng, Yan-Pei Cao, Ying Shan ARC Lab, Tencent PCG whcheng@tencent.com, caoyanpei@gmail.com, yingsshan@tencent.com
Pseudocode No The paper describes the system architecture and training/inference procedures in text and diagrams, but does not provide structured pseudocode or algorithm blocks.
Open Source Code No The paper does not contain an explicit statement of code release for the described methodology or a link to a code repository. The provided arXiv link is for the paper itself.
Open Datasets Yes We use the Scan Net dataset (Dai et al. 2017), following the original train/test split, to study the proposed and baseline methods.
Dataset Splits No The paper mentions following 'the original train/test split' of the ScanNet dataset and holding out specific scenes for testing, but it does not provide explicit details (percentages, counts, or a formal description) for a validation set split.
Hardware Specification Yes The experiments are conducted on NVIDIA V100 GPUs.
Software Dependencies No The paper mentions various models and optimizers used (e.g., Point-Ne RF, VQ-GAN, Adam optimizer) but does not provide specific version numbers for any software, programming languages, or libraries used in the implementation.
Experiment Setup Yes We train the model with a learning rate 1e-4 and batch size 16 using the Adam optimizer.