Skip to content

v1.2.0 — Vision node families

Latest
Latest

Choose a tag to compare

View all tags
@virgilvox virgilvox released this 24 Jun 04:31
v1.2.0
67aa27e
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

LATCH v1.2.0 — Vision node families

Adds 12 nodes across three families plus supporting infrastructure.

Nodes

  • snapshot — latch/hold a still frame from any texture feed (trigger or continuous).
  • object-detection-live — transformers.js YOLOS/DETR on a live feed, annotated-texture output.
  • object-detection-yolo — raw YOLOv8/v9 (GELAN) ONNX via onnxruntime-web; layout-robust parse + NMS.
  • OpenCV.js — new OpenCVService (lazy CDN load) + 9 nodes: grayscale, canny, threshold, blur, morphology, contours, corners, optical-flow, background-subtraction.

Verification

Real-browser proofs: OpenCV.js under credentialless COEP, YOLO detection on bus.jpg ({bus:1, person:4}), in-app shader → snapshot → main-output paint. 1481 unit tests + production build green.

Downloads

  • macOS (Apple Silicon / arm64) — LATCH-mac-arm64.dmg or LATCH-mac-arm64.zip
  • macOS (Intel / x64) — LATCH-mac-x64.dmg or LATCH-mac-x64.zip
  • Windows (x64) — LATCH-win.zip
  • Linux (x64) — LATCH-linux.deb or LATCH-linux.zip

The macOS builds are signed and notarized by Apple — they launch normally, no Gatekeeper workaround needed. Both Apple Silicon and Intel are included.

The web app is also deployed (GitHub Pages + Netlify).

Assets 11
Loading