Model
ReID + Gait
Re-identify subjects across cameras and time to build long-term dossiers.
shipped
Re-identification
IdentityEmbeddings
Overview
Appearance embeddings, soft biometrics (colour, height, accessories) and z-scored, winsorized gait descriptors are fused into a subject signature. Cosine matching links sightings across cameras and days, powering repeat-visitor detection, movement trails, relationship graphs and the identity dossier.
| Role | Re-identification |
|---|---|
| Expected latency | ~10–25 ms / embedding (GPU) |
| Hardware requirements | NVIDIA GPU recommended; embeddings serialised to avoid contention |
| Memory requirements | ~1 GB VRAM |
Inputs & Outputs
Inputs
- Person crop
- Track id
- Gait sequence
Outputs
- Appearance embedding
- Subject uid + match score
- Soft-biometric attributes
Advantages & Limitations
Advantages
- Cross-camera + cross-day identity
- Gait adds a hard-to-spoof cue
- Feeds dossiers and relationship graphs
Limitations
- Appearance drifts with clothing / lighting
- Gait needs a clean walking sequence
- Probabilistic, surfaced with confidence, not certainty
Example outputs
ReID + Gait
A repeat visitor flagged across three cameras
ReID + Gait
A long-term dossier with sighting histogram
ReID + Gait
A who-was-with-whom relationship graph
Configuration
Keys in config/default.yaml that govern this model.
| Key | Default |
|---|---|
roster.persist | true |
roster.persist_threshold | 0.74 |
gait.enabled | true |
Benchmarks
Placeholder
Benchmark figures below are illustrative and awaiting a standardised harness. Treat them as placeholders.
| Benchmark | Dataset | Metric | Value | Hardware |
|---|---|---|---|---|
| Rank-1 | internal | accuracy | placeholder | - |