Search papers, labs, and topics across Lattice.
This paper introduces a novel one-stage end-to-end model for fine-grained wildlife re-identification that integrates detection and identity searching within a latent space. By leveraging DINOv2 for spatial geometry and MegaDescriptor for re-identification, the model enhances latent queries with prompt features to effectively delineate object boundaries. Although the mean average precision score of 30.584% is lower than the state-of-the-art two-stage approach at 44.89%, qualitative results indicate promising advancements in bounding and identifying animal identities.
A one-stage model for wildlife re-identification challenges the traditional two-stage pipeline, achieving competitive results with enhanced latent querying.
Fine-grained wildlife re-identification remains a challenging area in research. Current state-of-the-art approaches apply a detection and re-identification pipeline. We propose a one-stage end-to-end detection and re-identification model that performs identity searching within the latent space. We adopt DINOv2 for robust spatial geometry and MegaDescriptor for wildlife re-identification. We enhance latent queries with prompt re-identification features. A detection decoder queries the scene latent space to establish object boundaries around the target identity. Preliminary findings reflect a competitive mean average precision score of 30.584% compared to the state-of-the-art two stage approach of 44.89%. Qualitative results depict effective bounding and identification of animal identities.
