SSS3D: Fast Neural Architecture Search For Efficient Three-Dimensional Semantic Segmentation

04/21/2023

∙

by   Olivier Therrien, et al.

∙

0

∙

share

We present SSS3D, a fast multi-objective NAS framework designed to find computationally efficient 3D semantic scene segmentation networks. It uses RandLA-Net, an off-the-shelf point-based network, as a super-network to enable weight sharing and reduce search time by 99.67 SSS3D has a complex search space composed of sampling and architectural parameters that can form 2.88 * 10^17 possible networks. To further reduce search time, SSS3D splits the complete search space and introduces a two-stage search that finds optimal subnetworks in 54 single-stage searches.