AI-Assisted Scene Mapping
Definition
Computer-vision models that automatically detect, tag and locate evidence items in bodycam, drone or scanner feeds, producing a draft evidence log for human review.
- Core technology
- Computer-vision detection models
- Input feeds
- Bodycam, drone, and scanner footage
- Output
- Draft evidence log for human review
- Role of human examiner
- Verifies and corrects the automated draft
Common questions
Why is human review still required if the detection is automated?+
Computer-vision models can misidentify objects, miss items outside their training distribution, or mislocate evidence in three-dimensional space, so an unreviewed draft log risks introducing errors into the chain of custody if entered as final documentation.
What advantage does this offer over an examiner logging evidence manually?+
It speeds up initial scene documentation and reduces the chance of an item being overlooked entirely during the walkthrough, since the system flags candidate objects across the whole feed for the examiner to confirm rather than relying purely on unaided visual scanning.
Related terms
- AR Walk-Through
- Augmented-reality overlay of evidence markers, measurements and notes onto a live or recorded view of the scene, viewable on tablet or headset.
- Bloodstain Pattern AI
- Trained models that classify bloodstain types (passive, transfer, impact, projected) and estimate impact angles from photographs, used as a triage aid before...
- Bodycam
- A body-worn camera, usually shoulder or chest mounted, used by the CSI to capture continuous video of scene work with timestamp and...
- Tele-Forensics
- Real-time video and data link between a crime-scene investigator and a remote forensic expert, allowing the expert to direct scene work as...
- VR Scene Reconstruction
- A fully immersive virtual model of the scene built from 3D scan and photogrammetry data, navigable in a headset for jurors, judges,...