The existing JIVISION multi-camera people-positioning project displays video, camera positions, person positions and trajectories in one interface. This review uses the published case and original project article to examine observation sources and association results. It adds no site tests. The original eight fixed network cameras were a reference configuration. Verification in that phase focused on video access, person observations and a display combining camera observations of one person.
Record date: 2026-09-11. Reference channel count: 8.
Check trajectory changes against original observations
The original article records logic using relative displacement between person observations from the same camera, consecutive observations and candidate-admission conditions. In saved stationary-observation replay, the adjusted logic suppressed position changes caused by observation-source changes. This is a record of the original replay scope. No new replay is reported. The original replay does not prove delay-free real motion. It does not prove that on-site position error has passed acceptance.
Separate display evidence and acceptance scope
| Item | Original evidence | Further verification needed |
|---|---|---|
| Live positions and trajectories | Display from the original single-person verification phase | Actual ground-position error and multiple people crossing |
| Sources and observation times | Camera sources, times and association states retained | Per-observation association details and error statistics |
| Stationary-observation replay | Suppression of source-change movement described in the article | Controlled routes with real motion, occlusion and reappearance |
| Captured face images | Related observation images displayed when conditions allowed | Face registration and identity-database comparison were not integrated in that stage |
Tools and curves do not replace complete evidence
The original article explicitly states that valid calibration of the whole site was unfinished and map positions were estimated coordinates. Site acceptance of absolute position error, multi-person cross-camera association and long-duration stability remained unfinished. Saved-observation replay does not establish acceptance of complete historical-trajectory search. This review has obtained no complete per-observation association table, new independent ground-position-error data or named-identity recognition results.
Preserve sources and unknown states in further checks
Further work can associate observations with original video segments, times, cameras, calibration versions and trajectories before and after processing. Check crossings, occlusion, source changes and real movement. Record unknown or pending-review states when evidence is insufficient. This is a proposed verification list, not a complete set of fields already documented for the original project. Eight cameras describe the original reference configuration, not the participating count in each result combining camera observations or a capacity limit.
Fact source: Multi-camera people-positioning case and the original project article. Related services: Multi-camera synchronization, calibration and fusion, project consultation.