DiffusionPointLabel: Annotated Point Cloud Generation with Diffusion Model

Authors: Li, T., Fu, Y., Han, X., Liang, H., Zhang, J.J. and Chang, J.

Journal: Computer Graphics Forum

Volume: 41

Issue: 7

Pages: 131-139

eISSN: 1467-8659

ISSN: 0167-7055

DOI: 10.1111/cgf.14663

Abstract:

Point cloud generation aims to synthesize point clouds that do not exist in supervised dataset. Generating a point cloud with certain semantic labels remains an under-explored problem. This paper proposes a formulation called DiffusionPointLabel, which completes point-label pair generation based on a DDPM generative model (Denoising Diffusion Probabilistic Model). Specifically, we use a point cloud diffusion generative model and aggregate the intermediate features of the generator. On top of this, we propose Feature Interpreter that transforms intermediate features into semantic labels. Furthermore, we employ an uncertainty measure to filter unqualified point-label pairs for a better quality of generated point cloud dataset. Coupling these two designs enables us to automatically generate annotated point clouds, especially when supervised point-labels pairs are scarce. Our method extends the application of point cloud generation models and surpasses state-of-the-art models.

https://eprints.bournemouth.ac.uk/40105/

Source: Scopus

DiffusionPointLabel: Annotated Point Cloud Generation with Diffusion Model

Authors: Li, T., Fu, Y., Han, X., Liang, H., Zhang, J.J. and Chang, J.

Journal: COMPUTER GRAPHICS FORUM

Volume: 41

Issue: 7

Pages: 131-139

eISSN: 1467-8659

ISSN: 0167-7055

DOI: 10.1111/cgf.14663

https://eprints.bournemouth.ac.uk/40105/

Source: Web of Science (Lite)

DiffusionPointLabel: Annotated Point Cloud Generation with Diffusion Model

Authors: Li, T., Fu, Y., Han, X., Liang, H., Zhang, J.J. and Chang, J.

Journal: Computer Graphics Forum

Volume: 41

Issue: 7

Pages: 131-139

ISSN: 0167-7055

Abstract:

Point cloud generation aims to synthesize point clouds that do not exist in supervised dataset. Generating a point cloud with certain semantic labels remains an under-explored problem. This paper proposes a formulation called DiffusionPointLabel, which completes point-label pair generation based on a DDPM generative model (Denoising Diffusion Probabilistic Model). Specifically, we use a point cloud diffusion generative model and aggregate the intermediate features of the generator. On top of this, we propose Feature Interpreter that transforms intermediate features into semantic labels. Furthermore, we employ an uncertainty measure to filter unqualified point-label pairs for a better quality of generated point cloud dataset. Coupling these two designs enables us to automatically generate annotated point clouds, especially when supervised point-labels pairs are scarce. Our method extends the application of point cloud generation models and surpasses state-of-the-art models.

https://eprints.bournemouth.ac.uk/40105/

Source: BURO EPrints