URVAM-ReID2026

URVAM-ReID2026 is a long-term urban object re-identification database designed for the evaluation of Re-ID systems in realistic city environments. This version of the dataset expands upon previous iterations by incorporating traffic signals into the established categories of crosswalks, trash bins, and containers. A significant feature of this edition is the inclusion of specific class labels for each category, enabling both re-identification and classification tasks. The dataset is composed of images extracted from four different video sequences capturing the same route in two directions through an urban environment. To ensure a truly representative long-term study, the recordings were conducted over a four-month period, capturing significant temporal, atmospheric, and environmental variations.