Luggage Re-identification
A fine-grained re-identification task for security scenarios such as airports and stations that tracks the same piece of luggage across surveillance video. Compared with pedestrian ReID, its target categories are finer, appearance differences are smaller, and occlusion is more frequent, requiring stronger feature discriminability and context modeling.
Timeline (last 90 days)
Existing baggage re-identification methods based on cropped image matching suffer significant performance degradation in real surveillance scenes due to occlusion, background clutter and crowded crowds.
Future work may need to further explore real-time optimization and validation on larger-scale real-world datasets.
All Facts (2)
Existing baggage re-identification methods based on cropped image matching suffer significant performance degradation in real surveillance scenes due to occlusion, background clutter and crowded crowds.
50%UnverifiedFuture work may need to further explore real-time optimization and validation on larger-scale real-world datasets.
50%