Work

Spatiotemporal Aggregation of Appearance and Motion Features for Video-based Person Re-Identification