Skip to content

v1.1.0: vismatch integration & cross-modal matching

Choose a tag to compare

View all tags
@0kam 0kam released this 17 Feb 05:02
· 6 commits to main since this release
v1.1.0
6fb9d77

What's New

vismatch integration (breaking change from internal API)

  • Migrated from imm (git+ URL) to vismatch (PyPI package)
  • Now installable via pip install alproj / uv add alproj (first PyPI release!)
  • All 70+ matching methods from vismatch are now available — no more hardcoded method list

New cross-modal matching methods

  • MatchAnything-RoMa: ~23 sec, 4999 matches — works on CPU (unlike minima-roma)
  • MatchAnything-ELoFTR: ~2 sec, 2547 matches — fast and robust
  • GIM-DKM: ~34 sec, 2048 matches — strong generalization from internet videos

Installation

# Basic
pip install alproj

# With deep learning matching
pip install "alproj[vismatch]"

Migration from v1.0.x

The optional dependency extra name changed from imm to vismatch:

-pip install "alproj[imm]"
+pip install "alproj[vismatch]"

The image_match() function API is unchanged. The internal IMM_METHODS list has been removed — any vismatch-supported method name is now accepted.

Full Changelog: v1.0.1...v1.1.0

Assets 2
Loading