yolov9 detect_dual.py was modified for use in an object and fire detection application that provides email and SMS alerts

[audioclassify]

The application runs on a low performance (and inexpensive) Ubuntu 22.04.4 desktop computer and provides alerts by email and SMS via a SMTP TLS or SSL mail server.

The CPU only inference times on a low performance desktop computer are between 350 and 400 mSec for Yolov9 object detection and 350 to 400 mSec for YOLOv9 fire detection.

Two weights (models) were trained using 'train_dual.py'

• yolov9-s-converted.pt
• yolov9-s-fire-converted.pt (using fire data from yolov5-fire-detection)

'detect_dual.py' and 'train_dual.py' are from WongKinYiu/yolov9.

For more information please see audioclassify/CedarAlert.