Real-Time Safety Detection System

I need a Python-based application that performs two main functions: social distance monitoring and thief detection in standard definition video.

Social Distance Monitoring:
- Detect individuals in a video stream using SSD MobileNet (Caffe model).
- Use depth estimation to calculate distances in real-time.
- Trigger a visual alert if someone is within 175 cm.

Thief Detection:
- Use YOLOv5 to detect individuals regularly.
- Implement OpenCV trackers to follow detected individuals.
- Monitor if a person lingers over five seconds in a sensitive area (ROI).
- Display a "THIEF DETECTED!" visual alert if this occurs.

Implementation Details:
- Load models using PyTorch (through torch.hub) and OpenCV DNN.
- Support real-time performance with frame skipping and tracker updates.
- Capture video input using imutils and VideoStream from the webcam.

Ideal Skills and Experience:
- Proficiency in Python
- Experience with computer vision libraries (OpenCV, PyTorch)
- Familiarity with object detection models (YOLOv5, SSD MobileNet)
- Knowledge in real-time video processing and alert systems