Saeejith Nair

Dec 1 2023, 11:30 am, EC4-2101A

*** 11:30am Presenter:  Saeejith Nair ***

Title: Optimizing Neural Radiance Fields: Towards More Tailored, Efficient View Synthesis

Neural radiance fields (NeRFs) enable high-quality novel view synthesis, but their high computational complexity limits deployability. While existing neural-based solutions strive for efficiency, they use one-size-fits-all architectures regardless of scene complexity. The same architecture may be unnecessarily large for simple scenes but insufficient for complex ones. Thus, there is a need to dynamically optimize the neural network component of NeRFs to achieve a balance between computational complexity and specific targets for synthesis quality. This talk will present our work on NAS-NeRF, a generative neural architecture search strategy that generates compact, scene-specialized NeRF architectures by balancing architecture complexity and target synthesis quality metrics. Our method incorporates constraints on target metrics and budgets to guide the search towards architectures tailored for each scene. Experiments on the Blender synthetic dataset show the proposed NAS-NeRF can generate architectures up to 5.74× smaller, with 4.19× fewer FLOPs, and 1.93× faster on a GPU than baseline NeRFs, without suffering a drop in SSIM. We will also discuss ongoing work on Elastic-NeRFs, a complementary approach that enables compact elastic deployment. By applying our elastic training strategy to existing NeRF methods, we show that we can yield multiple NeRF architectures at lower complexity that have similar representation capability as the original NeRF architecture