A Soft Fusion Approach for Graph Construction in Pedestrian Re-identification

As shown in Figure 1 and Figure 2, the lack of real pedestrian ID tags and the irregularity of pedestrian spatiotemporal migration often result in unreliable spatiotemporal scores. Therefore, the hard fusion method cannot accurately determine whether the images belong to the same pedestrian. The key is to reduce the noise caused by the fusion of spatiotemporal information. Based on this, we propose a soft fusion method for constructing graphs by selecting nodes with higher similarity based on spatiotemporal information within a high range of visual similarity and constructing the graph accordingly. The representation of the graph is as follows.