jbaruch/vision-pipeline-foundations
Hygiene patterns for any OpenCV + dlib vision pipeline: camera index probing + macOS init quirks, warmup that verifies real frames, frame-skip policy for expensive inference.
96
1.36x
Quality
93%
Does it follow best practices?
Impact
100%
1.36xAverage score across 6 eval scenarios
Securityby
Passed
No known issues
README.md
vision-pipeline-foundations
A Tessl plugin of foundation-layer patterns for any OpenCV + dlib CV pipeline. Not about a specific detector or classifier — about the hygiene you need below them.
Skills
| Skill | Purpose |
|---|---|
camera-setup | Reliable cv2.VideoCapture init with warmup, real-frame probe, macOS index-enumeration quirks. |
frame-skip-policy | Run expensive inference (dlib, ViT) at 1-of-N capture rate. Keeps the camera loop responsive. |
Install
tessl install jbaruch/vision-pipeline-foundationsWhy this matters
Nearly every "the pipeline isn't working" bug in a demo-facing CV app traces back to one of:
- The camera is opened but returning black frames (warmup).
- You opened index 0 which is a virtual camera, not the real one (probe).
- Inference is running on every frame and stalling the loop (frame-skip).
Encode these into a plugin so every new pipeline starts from the right place.
Composes with
face-recognition-calibration— producer-side CV model tuning, enrollment, persistenceiot-actuator-patterns— downstream actuator control with debounce + quantization
License
MIT — see LICENSE.