24,326 annotated images form the Real-Time Obstacle Detection (ROD) dataset. It was created at Amirkabir University of Technology, Tehran, to address sidewalk safety for pedestrians and people with visual impairments. The dataset page was last updated on June 7, 2026.
Use Cases
- Training YOLO-based obstacle detection models based on the 25-class annotation scheme
- Developing smartphone-based assistive vision systems based on the dataset's focus on sidewalk safety
- Benchmarking real-time object detection performance based on the dataset's size and urban context
- Researching pedestrian collision risk mitigation based on the dataset's public-safety problem statement
Strengths
- 24,326 images provide a substantial base for training computer vision models
- 25-class annotation scheme suggests detailed categorization of obstacles
- Dataset is the product of a university research project addressing a defined public-safety problem
Limitations
- Column-level documentation is absent; field semantics must be inferred after download
- Row count is unknown, which may limit suitability assessment
- Description metadata is limited; actual data quality requires manual inspection after download
Provenance
- Source
- Amirkabir University of Technology, Tehran
- Freshness
- Last updated 2026-06-07 03:30:09; freshness should be verified
- Geography
- Likely Tehran, Iran, based on the institution's location