Person Re-identification in Appearance Impaired Scenarios
This addresses a critical limitation in surveillance applications where traditional methods struggle, offering a solution for scenarios with impaired appearance, though it is incremental as it builds on existing re-identification frameworks.
The paper tackled the problem of person re-identification in scenarios where appearance-based features fail, such as when people wear similar clothing or change outfits, by proposing a dynamics-based feature that captures gait and motion patterns using Fisher vector encoded tracklets, and experiments on new datasets showed improved performance when combining dynamics with appearance information.
Person re-identification is critical in surveillance applications. Current approaches rely on appearance based features extracted from a single or multiple shots of the target and candidate matches. These approaches are at a disadvantage when trying to distinguish between candidates dressed in similar colors or when targets change their clothing. In this paper we propose a dynamics-based feature to overcome this limitation. The main idea is to capture soft biometrics from gait and motion patterns by gathering dense short trajectories (tracklets) which are Fisher vector encoded. To illustrate the merits of the proposed features we introduce three new "appearance-impaired" datasets. Our experiments on the original and the appearance impaired datasets demonstrate the benefits of incorporating dynamics-based information with appearance-based information to re-identification algorithms.