This is some Kodachrome level color voodoo – color grading and shot matching powered by machine-learning. And it comes from a collaboration with some friends of ours from the artist and live visual side, so it's doubly worth mentioning. What if the current techniques called AI turned out to be really important to creative artists – just not for the reason the general public expected? That's sure what Colourlab Ai looks like. It harnesses the powers of massive data crunching of pixels, the thing "AI" in the current generation was designed to do, and then applies it to making your video look amazing.
In a year when the holidays don't quite feel like the holidays -- what with COVID wreaking havoc on everyday lives and treasured traditions -- Netflix's Jingle Jangle: A Christmas Journey is here to serve up the extra dose of Christmas cheer so many of us are desperate for. And when I say extra, I do mean extra. Jingle Jangle, which hit Netflix Nov. 13, runs a perfectly reasonable two hours long, but those two hours are overstuffed in every way imaginable. There are too many concepts, too many subplots, and at least one framing device too many. The musical numbers are long, plentiful, and flashy.
We propose a new regularization method to alleviate over-fitting in deep neural networks. The key idea is utilizing randomly transformed training samples to regularize a set of sub-networks, which are originated by sampling the width of the original network, in the training process. As such, the proposed method introduces self-guided disturbances to the raw gradients of the network and therefore is termed as Gradient Augmentation (GradAug). We demonstrate that GradAug can help the network learn well-generalized and more diverse representations. Moreover, it is easy to implement and can be applied to various structures and applications. GradAug improves ResNet-50 to 78.79% on ImageNet classification, which is a new state-of-the-art accuracy.