Streamable Neural Fields
This addresses a practical bottleneck for users of neural fields in scenarios requiring efficient data streaming, though it is incremental as it builds on existing neural field paradigms.
The paper tackles the problem of data transfer inefficiency in neural fields by proposing streamable neural fields, a single model with executable sub-networks of various widths, enabling progressive reconstruction of signals; experimental results show effectiveness in domains like 2D images, videos, and 3D signed distance functions.
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.