The Devil is in the Details: Self-Supervised Attention for Vehicle Re-Identification

In recent years, the research community has approached the problem of vehicle\nre-identification (re-id) with attention-based models, specifically focusing on\nregions of a vehicle containing discriminative information. These re-id methods\nrely on expensive key-point labels, part annotations, and additional attributes\nincluding vehicle make, model, and color. Given the large number of vehicle\nre-id datasets with various levels of annotations, strongly-supervised methods\nare unable to scale across different domains. In this paper, we present\nSelf-supervised Attention for Vehicle Re-identification (SAVER), a novel\napproach to effectively learn vehicle-specific discriminative features. Through\nextensive experimentation, we show that SAVER improves upon the\nstate-of-the-art on challenging VeRi, VehicleID, Vehicle-1M and VERI-Wild\ndatasets.\n