An existing JIVISION multi-camera personnel-positioning project used eight fixed network cameras as its reference configuration, displaying video, channel states, observation sources and tracks in one interface. This review covers video inputs and source management using the existing project and the project manuscript dated 11 September 2026. It adds no new runtime tests or positioning-accuracy conclusions.
Video input is not positioning acceptance
The project continuously read video over the local network and retained times recorded by the image capture module and source-channel information. Personnel detection, tracking within a camera and single-target position fusion followed in the processing chain. Map positions also depended on site calibration, spatial relationships and observation selection. A working video display does not establish accepted position error or cross-camera identity association.
| Managed item | Information recorded in this project | Distinction to preserve |
|---|---|---|
| Video source | Fixed camera and channel | Source identifiers are not personnel identities |
| Image capture time | Timing information for an image observation | Image capture records are not automatically synchronized exposures |
| Channel state | Connected video and readiness | Readiness does not establish acceptance of every downstream function |
| Observation source | Cameras contributing to a target position | Contributing channels vary with position and occlusion |
| Displayed result | Staged video, position and track display | Display functions, absolute error and sustained operation need separate checks |
Relate observations to their sources
The original project retained source information between video and map views so an observed position could be checked against the relevant video. Fusion processing checked observation times, repeated detections and spatial relationships before selecting information. Source and timing fields also support investigation of missing, repeated or incorrectly associated observations.
These records do not establish synchronized exposure times across cameras. A source identifier is not a person's identity. Face images in the project records were for observation display; face registration and identity-database matching had not been integrated.
Limits of the recorded verification
Existing records describe staged verification of eight video inputs, personnel detection and single-target fusion display. Absolute positioning error, multi-person association across cameras and long-duration stability still require separate acceptance under the actual site conditions. The reference channel count does not establish unrestricted expansion or acceptance of other configurations.
Reusing existing cameras requires checks of stream access, mounting stability, image quality and ground coverage. Credentials and complete stream addresses are excluded from public project articles and managed under actual access rights and project agreements during handover.
Sources: multi-camera personnel-positioning project and JIVISION's project manuscript dated 11 September 2026. Related services: multi-camera synchronization, calibration and fusion and project enquiries.