Deyi Ji
Papers
1
Total Citations
2
H-Index
1
About
Deyi Ji is a rising star in computer vision whose research pushes the boundaries of 3D scene understanding by integrating large vision-language models. His primary focus lies in developing zero-shot and open-vocabulary methods for 3D perception, particularly in fine-grained reasoning and segmentation tasks. Ji’s most notable contribution is the introduction of "Reasoning3D," a groundbreaking framework that redefines 3D segmentation by enabling fine-grained, zero-shot, open-vocabulary part segmentation through natural language reasoning. This work transcends the limitations of traditional category-specific 3D semantic and instance segmentation, allowing models to locate and segment object parts based on complex, descriptive queries without any prior training on those specific categories. By leveraging large vision-language models, Ji’s approach bridges the gap between high-level reasoning and low-level 3D geometry, achieving a new paradigm for interactive 3D understanding. Although his seminal paper was published in 2024 and has already garnered 2 citations, its conceptual novelty and potential for applications in robotics, augmented reality, and autonomous systems mark him as a key innovator in the field. Ji’s work promises to make 3D AI more intuitive and accessible, setting a new standard for how machines interpret and interact with the three-dimensional world.
Research Focus
Key Achievements
Top Papers
- 1