FBPT: A Fully Binary Point Transformer

arXiv:2403.09998v1 Announce Type: new
Abstract: This paper presents a novel Fully Binary Point Cloud Transformer (FBPT) model which has the potential to be widely applied and expanded in the fields of robotics and mobile devices. By compressing the weights and activations of a 32-bit full-precision network to 1-bit binary values, the proposed binary point cloud Transformer network significantly reduces the storage footprint and computational resource requirements of neural network models for point cloud processing tasks, compared to full-precision point cloud …

abstract arxiv binary cloud cs.ai cs.cv devices fields mobile mobile devices network novel paper precision robotics transformer type values