Multi-library coverage (imgevolve is not HALCON-only)
Installed-library inventories introspected (ground truth, not scraped);
registry references counted via the cross-library catalog.
| library |
referenced by registry |
in installed inventory |
matched |
| OpenCV (cv2) |
98 |
484 public callables |
44 |
| scikit-image |
132 |
316 submodule functions |
84 |
Registry ops by source library (10+ libraries incorporated)
| library |
ops |
| core (numpy/scipy) |
306 |
| scikit-image |
86 |
| OpenCV |
50 |
| SimpleITK |
14 |
| Pillow |
13 |
| kornia (GPU) |
12 |
| scipy |
11 |
| mahotas |
10 |
| PyWavelets |
9 |
| scipy (3-D) |
8 |
| torch |
2 |
| total |
521 |
Honest reading
- Inventories are ALL public callables (cv2 ~484, skimage ~316); most are not
image-transform operators (IO, drawing, math, GUI, ML), so raw ratios understate reach.
- imgevolve wraps ecosystems rather than reimplementing them; adding an op with its
catalog analogue extends this automatically. Distinctive incorporations live in
backends_extra.py (xsk_/xcv_): inpainting, blob detectors, keypoint counts,
random-walker/flood/grabCut segmentation, structure/Hessian tensors, NPR filters.