Denis Shiryaev specializes in using AI-powered neural networks to upscale historical footage, and his latest work is this short clip of Moscow, Tverskaya Street in 1896 that was originally captured by Charles Moisson from the Lumière company. The neural networks were used upscale to 60FPS, boost image resolution with ESRGAN (general dataset), resort video sharpness, remove blur, remove compression artifacts, and colorization. Read more for the video and additional information.
Neural networks are basically networks of artificial neurons, or mathematical functions that transform a set of input values into an output value. The key feature is that they can be trained, so if you have lots of example inputs whose “correct” outputs are known, you can tune the parameters of the network to make it more likely to produce correct answers. The goal is that this training will generalize, in which once you’ve trained the neural networks to produce the right answer for inputs the network has seen before, it will produce good answers for inputs it hasn’t seen as well.
