HaCI - Bio-inspired distributed edge detection for RGB images
Show HN: My compact, bio inspired distributed edge detection algorithm
HaCI (Harmonic Contour Integration) is a compact, fully distributed algorithm that extracts clean edges from RGB images in four biologically inspired stages: pixel-level harmonic contrast, oriented pooling, seed facilitation and suppression, and ridge back-projection. It bridges gaps along contours while damping texture and gratings, producing sharp boundary maps. The bundled pretrained model scores 0.7179 ODS on the BRIND test set and runs in 36 ms at 481×321 on Apple M4. Install with uv, then run inference or evaluation out of the box. A paper is published on Zenodo.
Parallel contours on both sides of a cell divide it further, which damps texture and gratings but leaves a lone edge, or an edge beside a single parallel one, untouched.