Obstacles Avoidance Assistance for Visually Impaired

This dataset was curated to support research on obstacle avoidance systems for visually impaired individuals. It consists of annotated images representing four common obstacle categories encountered by pedestrians like pole, fence, bump, and hole. The dataset was compiled from publicly available sources on Roboflow and refined with careful manual labeling in the YOLO format. Each image includes bounding box annotations corresponding to the obstacle class. The images were collected under various lighting conditions (daylight, low-light, and shadows) and different outdoor environments such as pavements, roads, and pedestrian walkways.

The dataset contains a total of 1,627 annotations, distributed across classes as follows:
- Pole: 361 annotations
- Fence: 368 annotations
- Bump: 479 annotations
- Hole: 419 annotations