I use this a bit - the You Only Look Once implementation is good & fast (should be - redmon wrote the paper <a href="http://arxiv.org/abs/1506.02640" rel="nofollow">http://arxiv.org/abs/1506.02640</a>) and it's all C. I don't think it's competing to overtake Torch/Caffe/Tensorflow as 'the' framework. What I find it good for is looking at the source code and trying to understand how something works in code (often not easy even after reading the papers) - easier than with the big frameworks. Also may be a good place to start for simpler target platforms e.g. embedded. Kudos to pjreddie.
Bad name. Darknet already has a meaning.<p><a href="https://en.m.wikipedia.org/wiki/Darknet" rel="nofollow">https://en.m.wikipedia.org/wiki/Darknet</a><p>And why CUDA, and not OpenCL?
Don't get me wrong, I do appreciate the effort that went into this and fully intend to check it out sooner or later, but lately the appearance of new deep learning frameworks is beginning to trigger flashbacks to this: <a href="http://notinventedhe.re/on/2015-5-19" rel="nofollow">http://notinventedhe.re/on/2015-5-19</a>
This is cool. The documentation is very entertaining, although not something you'd show to your boss. Looks like it implements a lot more than just neural networks.<p>Shameless plug: For a minimal neural network implementation in ANSI C, check out: <a href="https://github.com/codeplea/genann" rel="nofollow">https://github.com/codeplea/genann</a> Sometimes lack of features is a feature.
Hmm, is this really nice? I doesn't look like something I'd just recommend to colleagues, with the religious symbols, names like Darknet and Yolo, Nightmare and black magic. Are they trying to stay away from the corporate market?