Adriano 6db2086ead feat: pyramid_propagate - candidati top-level guidano full-res ...

Top-level ritorna top-K picchi locali invece di solo max. Fase full-res
valuta solo crop locali attorno ai picchi propagati (margine =
sf_top + spread + nms_radius/2) invece di scansionare intera scena.

Su scene 1920x1080 con pochi candidati: ~20-30% piu veloce mantenendo
identici match. Vantaggio cresce con scene piu grandi e meno candidati.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

2026-05-04 15:26:29 +02:00

..

web

feat: upload file nella cartella IMAGES_DIR

2026-04-24 14:45:16 +02:00

__init__.py

perf: Numba JIT kernel per score_by_shift (2.1x speedup)

2026-04-24 01:30:31 +02:00

_jit_kernels.py

perf: Fase 2 speed (3x baseline) - fuse JIT + LRU + sub-pixel lazy

2026-04-24 21:21:59 +00:00

auto_tune.py

perf: Fase 2 speed (3x baseline) - fuse JIT + LRU + sub-pixel lazy

2026-04-24 21:21:59 +00:00

gui.py

ui: layout fisso 1600x900 con pannelli SX/DX, scena scalata

2026-04-24 08:59:06 +02:00

line_matcher.py

feat: pyramid_propagate - candidati top-level guidano full-res

2026-05-04 15:26:29 +02:00

matcher.py

perf: piramide al training, refinement sub-step, multithreading

2026-04-24 01:22:56 +02:00