I built a real-time people counting system 🛒👥 using YOLOv8 and OpenCV. It detects and tracks people crossing a line to count entries, exits, and net flow 📊, perfect for retail stores, gyms, or events. Supports video files, webcams, and live streams, with clear visual overlays so you see the numbers in real time.

Helps businesses make data-driven decisions on staffing, layout, and visitor flow based on accurate, real-time analytics

🎬 Real-time People Detection: Uses YOLOv8 Large model for accurate person detection

📊 Flexible Line-Based Counting: Horizontal or vertical line orientation for any movement direction

📈 Net Flow Analysis: Shows honest flow metrics instead of misleading occupancy counts

🎥 Multiple Input Sources: Support for video files, webcam, and live footage

🔧 Optimized Performance: Fine-tuned confidence thresholds and tracking parameters

💾 Export Capability: Save processed video with counting overlays

✅ Automatic Setup: Downloads required YOLO models automatically

🎛️ Command Line Interface: Comprehensive argument system for full control

Important: This system tracks Entry/Exit Flow, not absolute occupancy:

✅ Accurate: "12 people entered, 11 people left during this period"

✅ Honest: "Net flow: +1 person increase"

❌ Not claiming: "There are exactly X people in the area"

This is perfect for understanding traffic patterns, peak hours, and directional flow!