🟢 ddw denoiser

easymode denoise --method ddw

A distilled DeepDeWedge network. Separate DDW teacher networks were trained on each of 48 half-split datasets and then distilled into a single student network, so that the final network is general and can be applied directly to raw tomograms. Trained with Warp's tomogram reconstruction method.

A tomogram of a human T cell (Jurkat), raw input (left) and denoised output (right).

See general denoisers for the full description and for the optional arguments to easymode denoise.