Deep Learning on Implicit Neural Representations of Shapes
This addresses the challenge of integrating INRs into deep learning workflows for 3D shape processing, which is incremental as it builds on existing INR methods.
The paper tackles the problem of feeding Implicit Neural Representations (INRs) of 3D shapes into deep learning pipelines by proposing inr2vec, a framework that computes compact latent representations for INRs in a single inference pass, enabling effective embedding and use in downstream tasks.
Implicit Neural Representations (INRs) have emerged in the last few years as a powerful tool to encode continuously a variety of different signals like images, videos, audio and 3D shapes. When applied to 3D shapes, INRs allow to overcome the fragmentation and shortcomings of the popular discrete representations used so far. Yet, considering that INRs consist in neural networks, it is not clear whether and how it may be possible to feed them into deep learning pipelines aimed at solving a downstream task. In this paper, we put forward this research problem and propose inr2vec, a framework that can compute a compact latent representation for an input INR in a single inference pass. We verify that inr2vec can embed effectively the 3D shapes represented by the input INRs and show how the produced embeddings can be fed into deep learning pipelines to solve several tasks by processing exclusively INRs.