Learning A Predictable And Generative Vector Representation For Objects

Abstract

What is a good vector representation of an object? We believe that it should be generative in 3D, in the sense that it can produce new 3D objects; as well as be predictable from 2D, in the sense that it can be perceived from 2D images. We propose a novel architecture, called the TL-embedding network, to learn an embedding space with these properties. The network consists of two components: (a) an autoencoder that ensures the representation is generative; and (b) a convolutional network that ensures the representation is predictable. This enables tackling a number of tasks including voxel prediction from 2D images and 3D model retrieval. Extensive experimental analysis demonstrates the usefulness and versatility of this embedding.

Stats

• citations327
• S2 citations—
• github stars0
• HF likes0
• heat score18.87
• arxiv keygirdhar2016learning

Related papers

• A Convolutional Architecture For 3D Model Embedding (2021)0.00
• Learning A Hierarchical Latent-variable Model Of 3D Shapes (2017)12.87
• Mask2cad: 3D Shape Prediction By Learning To Segment And Retrieve (2020)12.87
• Self-aware Vector Embeddings For Retrieval-augmented Generation: A Neuroscience-inspired Framework For Temporal, Confidence-weighted, And Relational Knowledge (2026)0.00
• View N-gram Network For 3D Object Retrieval (2019)13.05
• Adversarial Autoencoders For Compact Representations Of 3D Point Clouds (2018)14.06
• Cross-modal Discrete Representation Learning (2021)10.61
• Learning Deep Representation Of Multityped Objects And Tasks (2016)0.00