Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

DC3DO: Diffusion Classifier for 3D Objects

Nursena Koprucu, Meher Shashwat Nigam, Shicheng Xu, Biruk Abere, Gabriele Dominici, Andrew Rodriguez, Sharvaree Vadgama, Berfin Inal, Alberto Tono

TL;DR

The approach, the Diffusion Classifier for 3D Objects (DC3DO), enables zero-shot classification of 3D shapes without additional training, demonstrating superior multimodal reasoning over discriminative approaches.

Abstract

Inspired by Geoffrey Hinton emphasis on generative modeling, To recognize shapes, first learn to generate them, we explore the use of 3D diffusion models for object classification. Leveraging the density estimates from these models, our approach, the Diffusion Classifier for 3D Objects (DC3DO), enables zero-shot classification of 3D shapes without additional training. On average, our method achieves a 12.5 percent improvement compared to its multiview counterparts, demonstrating superior multimodal reasoning over discriminative approaches. DC3DO employs a class-conditional diffusion model trained on ShapeNet, and we run inferences on point clouds of chairs and cars. This work highlights the potential of generative models in 3D object classification.

DC3DO: Diffusion Classifier for 3D Objects

TL;DR

The approach, the Diffusion Classifier for 3D Objects (DC3DO), enables zero-shot classification of 3D shapes without additional training, demonstrating superior multimodal reasoning over discriminative approaches.

Abstract

Inspired by Geoffrey Hinton emphasis on generative modeling, To recognize shapes, first learn to generate them, we explore the use of 3D diffusion models for object classification. Leveraging the density estimates from these models, our approach, the Diffusion Classifier for 3D Objects (DC3DO), enables zero-shot classification of 3D shapes without additional training. On average, our method achieves a 12.5 percent improvement compared to its multiview counterparts, demonstrating superior multimodal reasoning over discriminative approaches. DC3DO employs a class-conditional diffusion model trained on ShapeNet, and we run inferences on point clouds of chairs and cars. This work highlights the potential of generative models in 3D object classification.
Paper Structure (20 sections, 7 equations, 3 figures, 4 tables)

This paper contains 20 sections, 7 equations, 3 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methodology
  4. Multi-View Diffusion Classifier (MVDC)
  5. Diffusion Classifier for 3D Objects (DC3DO)
  6. Integrating LION with Diffusion Classifiers
  7. Integration with Hierarchical Latent Space
  8. Integration with Diffusion Models
  9. Encoding
  10. Diffusion Process
  11. Denoising and Classification
  12. Text-Conditioned Diffusion
  13. Experimental Results
  14. MVDC - 2D Results
  15. DC3DO Inference
  16. ...and 5 more sections

Figures (3)

  • Figure 1: Dataset classes for classification. We performed 3D Classifications tests on cars, chairs, and airplanes. We used multi-view and point cloud representations only for chairs and cars.
  • Figure 2: Methods comparison. We extended the Diffusion Classifierdiffusion_classifier paper to a multi-view su15mvcnn settings and we compare with our DC3DO model, based on lion
  • Figure 3: Visual Comparison. comparison of best(a-c) and worst(d-f) performing cars with accuracy percentages