Safety Outdoor Navigation Egocentric (SONE) Dataset

The Safety Outdoor Navigation Egocentric (SONE) Dataset contains RGB image sequences collected from a forward-facing egocentric camera for outdoor pedestrian navigation. The dataset is designed for research on safe mobility, anomaly detection, and scene understanding for assistive technologies. It includes normal navigation sequences (for training) as well as sequences containing obstacles (for testing), transitions, and visually ambiguous environments. The dataset supports benchmarking for autoencoder-based anomaly detection, multimodal fusion, and trajectory safety assessment.