Official implementation of 'Point-Bind & Point-LLM: Aligning Point Cloud with Multi-modality for 3D Understanding, Generation, and Instruction Following'. Using Point-Bind, we introduce Point-LLM, the ...
Abstract: Compared with 2D thermal images, visualizing the temperature of objects with their corresponding 3D surfaces provides a more intuitive way to perceive the environment. In this paper, we ...
Abstract: Semantic segmentation methods based on deep neural networks have achieved great success in recent years. However, training such deep neural networks relies heavily on a large number of ...