HolmeSketcher: Generative 3D Sketch Mapping for Spatial Reconstruction in Crime Scene Investigation
Tianyi Xiao, Yizi Chen, Sidi Wu, Peter Kiefer, Yan Feng, Martin Raubal

TL;DR
HolmeSketcher is a 3D sketch mapping system that enhances spatial reconstruction in crime scene investigation using deep learning and extended reality, improving accuracy but increasing task load.
Contribution
It introduces a novel 3D sketch mapping system combining a drawing interface with deep learning for scene reconstruction in CSI.
Findings
Improved spatial accuracy and interpretability of reconstructed scenes.
Higher task load and lower usability compared to 2D paper sketches.
Derived design implications for future 3D sketch mapping tools.
Abstract
Sketch mapping is widely used in crime scene investigation (CSI) to document, interpret, and communicate spatial information. However, it is typically performed on 2D media, which limits its ability to represent 3D spatial relationships. We present HolmeSketcher, a generative 3D sketch mapping system that combines a front-end 3D drawing interface with a back-end deep learning pipeline to support object generation and scene reconstruction in extended reality. In a within-subject user study (N = 15), HolmeSketcher improved the spatial accuracy and interpretability of reconstructed scenes, but with a clear trade-off of higher task load and lower usability compared with paper-based 2D sketch mapping. By integrating findings from the user study and expert interviews (N = 3), we further derive three design implications for next-generation 3D sketch mapping tools for CSI.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
