JCAD&CG Paper Reproducibility
Standardizes experimental reproducibility and artifact packages for graphics and CAD research papers submitted to JCAD&CG.
The JCAD&CG Paper Reproducibility skill assists researchers in packaging, verifying, and documenting experiments for the Journal of Computer-Aided Design & Computer Graphics (计算机辅助设计与图形学学报). It addresses graphics- and CAD-specific reproducibility challenges by establishing strict protocols for random seed fixing, GPU/driver and renderer version locking, mesh/point-cloud dataset hosting, rendering pipeline parameter alignment, clean-environment smoke testing, and evaluation metric calculation scripts.
Características Principales
01Structures standardized reproduction packages with one-click execution scripts
02Establishes reliable data and pre-trained model hosting guidelines using DOI archives
03Provides checklists for geometry (Chamfer/Hausdorff) and rendering (PSNR/SSIM/LPIPS) metric alignment
04Locks graphics-specific environmental parameters including GPU drivers, CUDA, APIs, and random seeds
051,079 GitHub stars
06Guides clean-environment smoke testing and tolerance threshold definitions for visual and numeric outputs
Casos de Uso
01Verifying rendering and geometric metric calculation consistency across different hardware setups
02Preparing clean code and dataset reproduction packages before submitting papers to JCAD&CG
03Responding to peer review feedback requiring artifact verification and experiment reproduction