Skip to content

Add Seen object detection to akiki.vision - #13

Merged
tobyrjackson merged 1 commit into
mainfrom
feat/vision-object-detection
Nov 18, 2026
Merged

tobyrjackson merged 1 commit into
mainfrom
feat/vision-object-detection

Conversation

@tobyrjackson

Copy link
Copy Markdown
Contributor

Summary

  • Add akiki.vision.object_detection.ObjectDetector and its Detection result type.
  • Load Seen r4c from the local Hugging Face cache or a supplied snapshot. Run RGB image detection through the local Akiki Seen engine.
  • Include the vision namespace in package discovery and distribution verification.

Verification

  • Ruff lint and formatting passed. Offline pytest passed 461 tests with 11 optional skips.
  • The cached Seen checkpoint returned the expected labels and boxes on a reference image.
  • On a Ryzen 9 9900X, the Python API and direct native API both took 3.00 ms median from RGB input to detections. Across 1,500 paired calls, median Python overhead was 5 µs.

@tobyrjackson
tobyrjackson merged commit 98ea5f3 into main Nov 18, 2026
4 checks passed
@tobyrjackson
tobyrjackson deleted the feat/vision-object-detection branch November 18, 2026 12:47
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant