John Bronskill

Papers

Total Citations

H-Index

About

John Bronskill is a leading researcher in few-shot learning and real-world object recognition, with a focus on enabling machines to learn new visual concepts from just a handful of examples—a critical capability for robotics, user personalization, and interactive AI. His most impactful contribution is the creation of the ORBIT dataset, a real-world few-shot benchmark designed for teachable object recognition. Unlike traditional datasets that rely on thousands of high-quality, curated images, ORBIT captures the messy, variable conditions of everyday use, including clutter, occlusion, and poor lighting, making it a rigorous test for practical AI systems. The dataset and accompanying paper have garnered over 35 citations, reflecting its growing influence in the computer vision and machine learning communities. Bronskill’s work bridges the gap between academic research and deployable technology, demonstrating how few-shot learning can be made robust for real-world applications. His contributions are particularly notable for advancing the paradigm of "teachable" AI—systems that can be quickly adapted by non-expert users. For students and researchers, Bronskill’s research offers a compelling blueprint for building more flexible, human-centric recognition systems.