4 ms·They're frames extracted from video as a convenient way to get a large enough training set (50000 images).by Luc 10y agoThey're frames extracted from video as a convenient way to get a large enough training set (50000 images).Jaruzel 10y agoIt's probably still worth running them all though a de-noise filter first tho.pjc50 10y agoAny half-decent "deep learning" system should be noise-tolerant.