New ORBIT‑India dataset targets blind users in India The dataset extends personalized computer‑vision tools to India, home to the world’s largest blind population, addressing gaps where existing systems trained on high‑resource data fail to recognize everyday objects for blind individuals [1].
Dataset contains over 105,000 images of 76 objects Collected from 587 videos, the collection provides a substantial visual resource for training AI models to recognize personal items such as keys, notebooks, and canes in Indian material culture [1].
Study examines challenges of low‑resource data collection Researchers detail how practices from high‑resource settings must be adapted for India, highlighting issues like limited equipment, diverse lighting, and cultural object variations, and they propose recommendations for cross‑geography dataset creation [1].
Findings presented at CHI 2026 conference The work was peer‑reviewed and presented at the CHI Conference on Human Factors in Computing Systems, underscoring its relevance to the HCI community and future accessibility research [1].