MSP-MVS: Multi-Granularity Segmentation Prior Guided Multi-View Stereo

Authors: Zhenlong Yuan, Cong Liu, Fei Shen, Zhaoxin Li, Jinguo Luo, Tianlu Mao, Zhaoqi Wang

AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Evaluations on ETH3D and Tanks & Temples benchmarks prove our method obtains state-of-the-art performance with remarkable generalization. ... Experiment Datasets and Implementation Details We evaluate our work on the ETH3D benchmark (Schops et al. 2017) and the Tanks & Temples benchmark (TNT) (Knapitsch et al. 2017). ... Results on ETH3D and TNT ... Ablation Study
Researcher Affiliation Academia 1Institute of Computing Technology, Chinese Academy of Sciences 2Harbin Institute of Technology, Shenzhen 3Nanjing University of Science and Technology, Nanjing 4Agricultural Information Institute, Chinese Academy of Agricultural Sciences 5Key Laboratory of Agricultural Big Data, Ministry of Agriculture and Rural Affairs EMAIL, EMAIL, EMAIL, EMAIL, EMAIL, EMAIL
Pseudocode No The paper describes its methodology in text and mathematical equations (e.g., equations 1-9) and outlines components like 'Multi-Granularity Segmentation Prior', 'Attention-Consistent Patch Deformation', and 'Disparity-Sampling Synergistic 3D Optimization', but it does not include a clearly labeled pseudocode block or algorithm.
Open Source Code No The paper does not contain any explicit statement about releasing source code, a link to a code repository, or a mention of code being available in supplementary materials.
Open Datasets Yes We evaluate our work on the ETH3D benchmark (Schops et al. 2017) and the Tanks & Temples benchmark (TNT) (Knapitsch et al. 2017).
Dataset Splits Yes We evaluate our work on the ETH3D benchmark (Schops et al. 2017) and the Tanks & Temples benchmark (TNT) (Knapitsch et al. 2017). ... All experiments are performed on original-resolution images in both ETH3D dataset and TNT dataset. ... All experiments are performed on original-resolution images, with the number of images standardized to 10 across all scenarios for runtime evaluation.
Hardware Specification Yes Our method is implemented on a machine with an Intel(R) Xeon(R) Silver 4210 CPU and eight NVIDIA Ge Force RTX 3090 GPUs.
Software Dependencies No The paper does not explicitly list any specific software dependencies or their version numbers, such as programming languages, libraries, or frameworks used for implementation.
Experiment Setup Yes In experiment, w = 11, λ = 0.25, θ = 45 , |S| = 8. ... We take APD-MVS (Wang et al. 2023) as our baseline. ... Table 3: Quantitative ablation studies on ETH3D dataset. ... w/o. Cst. ... µ = 3 ... MSP-MVS ... µ = 5