Installation & quickstart

This guide walks you through installing vigilo-stream, setting up ONNX model weights, and running your first vision pipeline.

Installation

Install the package from PyPI:

pip

pip install vigilo-stream

uv

uv add vigilo-stream

poetry

poetry add vigilo-stream

You can import the package using either vigilo_stream or the legacy alias rustream:

import vigilo_stream
# or
import rustream as vigilo_stream

System requirements

• Python: 3.9, 3.10, 3.11, 3.12, 3.13, 3.14, or 3.15.
• Operating systems:
  • Windows 10 / 11 (x86_64)
  • Ubuntu 20.04+ / Debian 11+ (x86_64)
  • macOS 12+ (Apple Silicon arm64 or Intel x86_64)
• Hardware acceleration (optional):
  • Integrated GPUs (iGPUs / APUs): Supported via DirectML. For real-time performance, Intel 11th Gen Core and newer (Iris Xe / Intel Arc) or AMD Ryzen 6000 series and newer (RDNA 2 / RDNA 3, e.g. Radeon 680M / 780M) are recommended. Older integrated graphics run best on the default CPU engine.
  • Discrete GPUs: NVIDIA GeForce/RTX, AMD Radeon RX, or Intel Arc.
  • Apple Silicon: M1, M2, M3, M4 (CoreML / Metal).
  • See full details in the Hardware & GPU acceleration guide.
• Optional: OpenCV (opencv-python or opencv-python-headless) for video display and HUD rendering.

Model weights

The neural pipeline uses four ONNX Runtime models:

Task	Architecture	Default filename	Size
Face detection	YuNet (320x320)	face_detection_yunet_2023mar.onnx	~230 KB
Head pose	MobileNetV3 Small	headpose_mobilenetv3_small.onnx	~6.1 MB
Gaze estimation	MobileOne-S0 Gaze	mobileone_s0_gaze.onnx	~5.0 MB
Object detection	YOLOX-Nano	yolox_nano.onnx	~3.7 MB

Automatic download

By default, Pipeline(models_dir="models", auto_download=True) checks for missing models and downloads them automatically on first use.

You can also download them explicitly ahead of time:

import vigilo_stream

# Downloads all missing models into the ./models directory
vigilo_stream.download_models("models")

Manual download

If you prefer to download weights manually:

mkdir -p models
curl -sSL -o models/face_detection_yunet_2023mar.onnx https://github.com/opencv/opencv_zoo/raw/main/models/face_detection_yunet/face_detection_yunet_2023mar.onnx
curl -sSL -o models/headpose_mobilenetv3_small.onnx https://github.com/yakhyo/head-pose-estimation/releases/download/weights/mobilenetv3_small.onnx
curl -sSL -o models/mobileone_s0_gaze.onnx https://github.com/yakhyo/gaze-estimation/releases/download/weights/mobileone_s0_gaze.onnx
curl -sSL -o models/yolox_nano.onnx https://github.com/Megvii-BaseDetection/YOLOX/releases/download/0.1.1rc0/yolox_nano.onnx

Quickstart

Here is a minimal script running camera capture, neural detection, and temporal event polling:

import time
import numpy as np
import vigilo_stream

# 1. Initialize pipeline with auto-download
pipe = vigilo_stream.Pipeline(models_dir="models", auto_download=True)
pipe.start("camera:0")

print("Pipeline started. Press Ctrl+C to stop.")

try:
    while pipe.is_running():
        # Poll the latest frame (zero-copy RGB8)
        frame = pipe.poll_frame()
        if frame is not None:
            # Direct pointer exposure into NumPy array
            img = np.asarray(frame)
            print(f"Frame seq={frame.seq} shape={img.shape}", end="\r")

        # Poll latest detection signals
        snapshot = pipe.snapshot()
        if snapshot and snapshot.faces:
            face = snapshot.faces[0]
            print(f"\nFace detected! Score: {face.score:.2f} BBox: {face.bbox}")

        # Drain new temporal violation events
        for event in pipe.events():
            print(f"\nAlert: {event.event_type} - {event.violation}")

        time.sleep(0.01)

except KeyboardInterrupt:
    print("\nStopping...")

finally:
    pipe.stop()
    print("Pipeline stopped.")