A Dataset of Non-Invasive Deep Learning Methods for Imaging through Dynamic Scattering Media

The research results of this paper are analyzed from this experimental data set. Task Description: This paper mainly solves the imaging problem in actual scattering scenarios. The data was collected under real scattering experimental conditions and can be used for training neural networks.

It includes experimental data on different degrees of scattering, with clear scattering images at low concentrations and almost indistinguishable scattering images at high concentrations. Low and high concentration data form a set of training pairs that can guide network learning from scattering maps to clear maps.