@article{ART003383270},
author={Jaehong Kim and Jaeseok Seo and Kiwook Park and KIM KWANG SOO and Lee, Ki Hoon},
title={Comparative Analysis of XAI Performance in 3D CNN Models with Different Point Cloud Voxelization Methods},
journal={ Journal of Software Forensics},
issn={3092-541X},
year={2026},
volume={22},
number={3},
pages={205-215}
TY - JOUR
AU - Jaehong Kim
AU - Jaeseok Seo
AU - Kiwook Park
AU - KIM KWANG SOO
AU - Lee, Ki Hoon
TI - Comparative Analysis of XAI Performance in 3D CNN Models with Different Point Cloud Voxelization Methods
JO - Journal of Software Forensics
PY - 2026
VL - 22
IS - 3
PB - Korea Software Assessment and Valuation Society
SP - 205
EP - 215
SN - 3092-541X
AB - This paper analyzes input channel contributions and spatial activations of a dense 3D CNN for urban point cloud semantic segmentation from an Explainable AI (XAI) perspective. A cropped region of the STPLS3D dataset was voxelized using the Full-Z+RGB-16 representation, consisting of occupancy, normalized relative z-height, R16, G16, and B16, and GradientSHAP, Grad-CAM, and ROAR were applied.
The results showed that occupancy had the highest mean absolute SHAP attribution for both Building and Car, while the relative importance of height and color channels differed by class. Grad- CAM activations were observed around occupied voxels as well as near patch boundaries, indicating a possible boundary effect. In the ROAR experiment, at a 40% channel removal rate, the Car IoU after SHAP ranked removal was 20.79%, approximately 15.45 percentage points lower than the 36.24% obtained with random removal.
KW - 3D CNN;Voxelization;Explainable AI (XAI);Semantic Segmentation
DO -
UR -
ER -
Jaehong Kim, Jaeseok Seo, Kiwook Park, KIM KWANG SOO and Lee, Ki Hoon. (2026). Comparative Analysis of XAI Performance in 3D CNN Models with Different Point Cloud Voxelization Methods. Journal of Software Forensics, 22(3), 205-215.
Jaehong Kim, Jaeseok Seo, Kiwook Park, KIM KWANG SOO and Lee, Ki Hoon. 2026, "Comparative Analysis of XAI Performance in 3D CNN Models with Different Point Cloud Voxelization Methods", Journal of Software Forensics, vol.22, no.3 pp.205-215.
Jaehong Kim, Jaeseok Seo, Kiwook Park, KIM KWANG SOO, Lee, Ki Hoon "Comparative Analysis of XAI Performance in 3D CNN Models with Different Point Cloud Voxelization Methods" Journal of Software Forensics 22.3 pp.205-215 (2026) : 205.
Jaehong Kim, Jaeseok Seo, Kiwook Park, KIM KWANG SOO, Lee, Ki Hoon. Comparative Analysis of XAI Performance in 3D CNN Models with Different Point Cloud Voxelization Methods. 2026; 22(3), 205-215.
Jaehong Kim, Jaeseok Seo, Kiwook Park, KIM KWANG SOO and Lee, Ki Hoon. "Comparative Analysis of XAI Performance in 3D CNN Models with Different Point Cloud Voxelization Methods" Journal of Software Forensics 22, no.3 (2026) : 205-215.
Jaehong Kim; Jaeseok Seo; Kiwook Park; KIM KWANG SOO; Lee, Ki Hoon. Comparative Analysis of XAI Performance in 3D CNN Models with Different Point Cloud Voxelization Methods. Journal of Software Forensics, 22(3), 205-215.
Jaehong Kim; Jaeseok Seo; Kiwook Park; KIM KWANG SOO; Lee, Ki Hoon. Comparative Analysis of XAI Performance in 3D CNN Models with Different Point Cloud Voxelization Methods. Journal of Software Forensics. 2026; 22(3) 205-215.
Jaehong Kim, Jaeseok Seo, Kiwook Park, KIM KWANG SOO, Lee, Ki Hoon. Comparative Analysis of XAI Performance in 3D CNN Models with Different Point Cloud Voxelization Methods. 2026; 22(3), 205-215.
Jaehong Kim, Jaeseok Seo, Kiwook Park, KIM KWANG SOO and Lee, Ki Hoon. "Comparative Analysis of XAI Performance in 3D CNN Models with Different Point Cloud Voxelization Methods" Journal of Software Forensics 22, no.3 (2026) : 205-215.
