Group Detection with Entry–Exit Re-identification
Description
Identifies groups of visitors and calculates their dwell time within a zone by re-identifying people as they enter and exit as a group.
This method combines group detection at access points with entry–exit re-identification, enabling the measurement of dwell time and conversion metrics at a group level.
| Element | Detail |
|---|---|
| Description | Identifies visitor groups at entry and re-identifies them at exit to calculate group dwell time. |
| Algorithm | 1. Group entries. 2. Detect groups based on temporal and physical proximity. 3. Group exits. 4. Re-identify entry–exit groups by temporal similarity and cardinality. 5. Calculate group dwell time. |
| Key parameters | • Line perimeter (defines the zone) • Re-ID window (default 2 h) • Temporal grouping window (1-3 s) |
| Accuracy | ~95% on validated datasets. Accuracy may decrease at high-density entrances (>600 people/hour by entrance). |
| Typical use cases | • Group conversion analysis • Family and social behavior analysis • Layout and staffing optimization |
Visual Explanation
| Visual explanation |  |  |  |
|---|---|---|---|
| Description | 1. Entry grouping: people crossing the entry perimeter are detected. | 2. Exit grouping: people leaving the zone are detected. | 3. Temporal grouping and group detection: people are grouped based on temporal proximity to form groups, which are then re-identified between entry and exit, enabling group dwell time calculation. |