Model weights & helpers

Utilities for managing, downloading, and verifying ONNX model files.

Helper functions

download_models(target_dir="models")

Downloads all missing ONNX model weights from their official upstream repositories.

• Parameters:
  • target_dir (str, optional): Target directory where model files are saved. Created automatically if it does not exist. Defaults to "models".
• Returns: None
• Raises:
  • RuntimeError: If any model download fails or times out.

import vigilo_stream

# Download all missing models into the default ./models directory
vigilo_stream.download_models()

# Download into a custom path
vigilo_stream.download_models("/opt/models")

---

Constants

MODEL_URLS

A dictionary mapping each model filename to its official download URL:

import vigilo_stream

for filename, url in vigilo_stream.MODEL_URLS.items():
    print(f"{filename} -> {url}")

Filename	Source repository
face_detection_yunet_2023mar.onnx	OpenCV Zoo YuNet
headpose_mobilenetv3_small.onnx	yakhyo/head-pose-estimation
mobileone_s0_gaze.onnx	yakhyo/gaze-estimation
yolox_nano.onnx	Megvii-BaseDetection/YOLOX
w600k_mbf.onnx	deepinsight/insightface