Streamable Neural Fields

TL;DR

Streamable neural fields introduce a single adaptable model composed of sub-networks of varying widths, enabling progressive signal reconstruction and efficient data transfer, applicable across multiple domains like images, videos, and 3D models.

Contribution

The paper presents a novel architecture and training method for neural fields that allows a single model to be streamable over time with multiple quality levels, enhancing practicality and stability.

Findings

Effective in 2D images, videos, and 3D signed distance functions

Enables progressive signal reconstruction with varying quality

Improves training stability through parameter sharing

Abstract

Neural fields have emerged as a new data representation paradigm and have shown remarkable success in various signal representations. Since they preserve signals in their network parameters, the data transfer by sending and receiving the entire model parameters prevents this emerging technology from being used in many practical scenarios. We propose streamable neural fields, a single model that consists of executable sub-networks of various widths. The proposed architectural and training techniques enable a single network to be streamable over time and reconstruct different qualities and parts of signals. For example, a smaller sub-network produces smooth and low-frequency signals, while a larger sub-network can represent fine details. Experimental results have shown the effectiveness of our method in various domains, such as 2D images, videos, and 3D signed distance functions. Finally, we demonstrate that our proposed method improves training stability, by exploiting parameter sharing.