Attention and Multi-granied Feature Learning for Baggage Re-identification

Abstract

AbstractThe current baggage re-identification methods only consider the global coarse-grained features while ignoring the fine-grained features. To deal with this issue, we proposed a simple and efficient multi-granularity feature learning based on attention mechanisms for this task. First, we introduce the global context attention mechanism into the backbone network to improve global features. Then, we use the batch feature dropbox module to learn local fine-grained features with context information combined with global coarse-grained feature learning. Our method achieved 84.6% Rank-1 and 82.5% mAP on the public dataset, which verified the performance of baggage re-identification can be improved by global context information and multi-granularity feature learning.KeywordsBaggage re-identificationAttention mechanismsMulti-grained feature learning