Duy-Dinh Le

Vietnam National University Ho Chi Minh City

Papers

Total Citations

H-Index

About

Duy-Dinh Le is a leading researcher in computer vision, with a focused expertise in object detection and the challenge of recognizing unseen or novel objects. His work addresses a fundamental limitation of traditional deep learning models, which typically fail when encountering objects not present in their training data. Le’s major contributions are encapsulated in two seminal papers: “You always look again: Learning to detect the unseen objects” (2019) and its follow-up, “YADA: you always dream again for better object detection” (2019). These works introduce innovative frameworks that enable models to dynamically “look again” or “dream” of potential objects, effectively learning to detect the unseen by leveraging contextual cues and iterative refinement. While these papers have garnered 10 and 8 citations respectively, their impact lies in pioneering a paradigm shift toward open-world object detection—a critical step for real-world applications like autonomous driving and robotics. Le’s research bridges the gap between closed-set and open-set recognition, inspiring subsequent work on continual learning and zero-shot detection. His contributions are particularly notable for their elegant, metaphor-driven approach to a hard problem, making complex concepts accessible and advancing the field’s ability to handle the unpredictable visual world.