3D segmentation studies without RGB-D depth focus on using solely RGB images or point cloud data for segmenting objects in three-dimensional space. Research works in this area often explore methods such as deep learning techniques that leverage convolutional neural networks (CNNs) on 2D projections, as well as utilizing geometric features from point clouds through architectures like PointNet or voxel-based representations to achieve accurate segmentation without relying on depth information.

Dear Tong Guo ,

3D segmentation studies without RGB-D depth focus on using solely RGB images or point cloud data for segmenting objects in three-dimensional space

https://arxiv.org/html/2103.05423v4

Regards,

Shafagat