Dynamic-excitation-based steady-state non-line-of-sight imaging via multi-branch convolutional neural network

•We construct a virtual rendering pipeline for generating two large-scale rendered steady-state NLOS image datasets under dynamic excitation setting.•We build a physical hardware setup for acquiring realistic steady-state NLOS images containing objects of various shapes and in different poses under dynamic excitation setting.•We propose a multi-branch feature fusion computational framework to process multiple NLOS images captured under a dynamic excitation setting to perform high-quality steady-state NLOS imaging tasks.