Skip to content

inference.md

Latest commit

 

History

History
57 lines (41 loc) · 1.79 KB

inference.md

File metadata and controls

57 lines (41 loc) · 1.79 KB

Inference Loop & Features

The implementation in src/inference_loop.py provides a robust detection loop with hardware integration.

Key Features

  1. Model Support: Runs both Legacy TFLite and Modern ONNX models via IREE runtime.
  2. Lighting Control: Synchronizes LED lighting with camera capture to ensure consistent image quality.
  3. Resolution Control: Optimizes local USB camera input for performance.
  4. Dry Run Mode: Simulutes the entire pipeline for safe testing without hardware.

Configuration

Control the behavior using environment variables in your .env file.

General

Variable Default Description
MODEL_PATH models/spaghetti_v2.vmfb Path to the compiled IREE module.
THRESHOLD 0.5 Confidence threshold for detection.
DRY_RUN false Set to true to simulate hardware interactions.

Camera & Resolution

Variable Default Description
CAMERA_TYPE http local (USB) or http (MJPEG stream).
CAMERA_WIDTH 640 Resolution width for local camera.
CAMERA_HEIGHT 480 Resolution height for local camera.

Lighting Control

Variable Default Description
LIGHTING_ENABLED false Enable/Disable lighting synchronization.
LIGHTING_GCODE_ON SET_LED... G-code to turn lights ON.
LIGHTING_GCODE_OFF SET_LED... G-code to turn lights OFF.

Testing with Dry Run

To verify the logic without connecting a printer or camera:

DRY_RUN=true \
CAMERA_TYPE=local \
LIGHTING_ENABLED=true \
MODEL_PATH=models/model.vmfb \
python3 src/inference_loop.py

This will:

  • Log G-code commands instead of sending them.
  • Generate mock images.
  • Use a mock model (no inference).
  • Log pause commands.