OF---OnlyFeet

An advanced computer vision system for real-time person counting on escalators using YOLOv8 Nano, OpenCV, zone-based detection, and advanced tracking techniques.

Real-Time Escalator Person Counter using YOLOv8 Nano & SORT

---

What It Does

• Detects and tracks people on escalators in real-time
• Uses YOLOv8 Nano for fast and efficient object detection
• Tracks individuals using the SORT algorithm
• Displays bounding boxes and ID labels for each tracked person

• Applies zone-based detection to focus only on specific escalator region for accurate count.

---

Key Technologies

• YOLOv8 Nano – Lightweight object detection
• SORT Tracker – Simple real-time object tracking
• OpenCV + CvZone – Image processing and video handling
• Zone-based ROI Masking – Only detect within defined region

---

How It Works

1. Load video and resize to 640×360
2. Apply region-based mask to filter non-escalator areas
3. Use YOLOv8 to detect only 'person' class
4. Use SORT to track each person across frames
5. Show unique ID and bounding box per person in real-time

---

Requirements

pip install ultralytics opencv-python numpy cvzone

add sort.py or use a pip-based SORT library.

---

Run the App

python yolo.py

Tested on 1080p escalator video input.

---

Use Cases

• Malls & Airports – Escalator usage monitoring
• Metro Stations – Real-time crowd flow analysis
• Smart Infrastructure – People counting systems

---

Project Structure

EscaCount/
├── person_counter.py
├── mask.png
├── video.mp4
├── requirements.txt
└── README.md

---

Made by Jainil

“Because real-time person tracking should be both smart and scalable.”

---