News for hackers. Stuff that matters.

RenderBear

Thanks! A few diagrams on the spatial layout and information flow between stages would indeed go a long way. I'll take that up in the upcoming commits.
I've added a few more to the README.

For a breakdown of the algo's stages on an image, the -d flag on inference generates detailed views like cell density, orientations, etc.

Show HN: My compact, bio inspired distributed edge detection algorithm

https://github.com/RenderBear/haci

Hi HN,I've been working on a bio / visual cortex inspired edge detection algorithm. I released a preliminary version a couple of months ago, and I've been improving it since. Benchmark scores currently sit midway between old-school (Canny, Sobel), and modern deep nets, while having the former's interpretability.I'm planning to develop this further for downstream use cases, maybe try to improve...

( Read More... | 4 comments | on HN )
Hi all,

I've developed an edge detection algorithm inspired by functions of the visual cortex such as divisive normalization and spatial pooling. This goal here is to sit in the class of compact methods like Canny, while having a small set of trainable parameters that adjust to data.