Go to content Go to navigation

Time flies · 2019-03-24 11:59 by Black in

After finishing my masters thesis and completing civil service, I’ve been working for Disney Research since early 2010. And I’ve obviously not had much to say on this blog. Most of my early work was focused on software engineering, mostly tech transfer using C++. I’ve written the basis of a big code base that is used (and considering the resistance to update anything will likely be used for decades) in many projector / camera systems in disney parks. Qt was popular for GUI work back then, and I’ve used it in several projects targeted at normal humans (but since those were internal research projects, not many actually used it). I’ve worked on projects targeting mobile platforms, in both Objective-C and Unity (I’m even named on a patent for one of them). I’ve written plugins for Nuke, ToonBoom, AfterEffects and many others, code running on arduinos or servers with dozens of cores and multiple GPUs.

But lately, I’ve shifted to projects that focus on machine learning. I’m not a researcher, so I don’t focus on developing models and graphs, but over the past few years I’ve debugged, ported and improved a lot of the deep learning research code created here. My favorite framework is TensorFlow, and I spent most time with it, but I’ve also used PyTorch. TensorFlow’s graph based structure forces more discipline, and makes reasoning much easier than the python spaghetti code I’ve seen from torch users. The biggest NN project I’ve participated in so far was Denoising with Kernel Prediction. Implemented in TensorFlow, and using custom CUDA code for better performance, this denoiser surpasses anything before.

I still like C++, it is one of the most flexible and pragmatic languages that are widely used. But for deep learning, Python is the standard, and inertia ensures this will remain the case for many years. Which is unfortunate, the lack of type safety and static checking is a big annoyance, especially in non-trivial code bases as the ones I work in.
Using C++ for TensorFlow is possible but only done for very specific subset of tasks, such as embedding into other programs, or writing custom Ops with CPU or CUDA.

  Textile help