[MateoLostanlen]

• The engine image installed requirements.txt with dependency resolution, so uv added the latest opencv-python (5.0.0.93) for ncnn, next to the pinned opencv-python-headless 4.11.
• Both packages write cv2/, so the install order decided which one was used. On some CPUs opencv-python 5.0 crashes on import cv2 (SIGILL), and the published latest image has it.
• Install requirements.txt with --no-deps, since it already pins every dependency. The only change in the installed packages is that opencv-python is gone.
• Add a CI check that fails if opencv-python is installed, and imports cv2.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 80.51%. Comparing base (a2604d4) to head (f214bf4).

Additional details and impacted files

@@           Coverage Diff            @@
##           develop     #460   +/-   ##
========================================
  Coverage    80.51%   80.51%           
========================================
  Files            6        6           
  Lines          847      847           
========================================
  Hits           682      682           
  Misses         165      165           

Flag	Coverage Δ
unittests	80.51% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.