An intelligent traffic light control system based on computer vision.
Analyzes the road situation in real time, detects vehicles and pedestrians,
calculates optimal traffic light phase durations, and saves statistics.

• region of Interest selection to count objects only within the road area
• object detection: cars, motorcycles, buses, trucks, pedestrians (YOLOv8)
• object tracking (ByteTrack)
• speed estimation
• elderly pedestrian detection (to increase crossing time)
• emergency vehicle recognition (priority switching)
• dynamic green time calculation for vehicles and pedestrians
• incident detection (sudden stop)
• traffic density heatmap
• real‑time density mini‑graphs
• environmental analytics (estimated CO₂ and fuel savings)
• CSV and JSON logging for dashboard
• photo analysis mode with annotated image and JSON report

1. Clone the repository:

   git clone https://github.com/IbrokhimN/trafic-detector.git
   cd trafic-detector

2. Install dependencies:

   pip install -r src/requirements.txt

   Note: The YOLOv8l model will be downloaded automatically on first run.

python src/main.py                # webcam
python src/main.py video.mp4      # video file
python src/main.py rtsp://user:pass@ip/stream   # RTSP stream

On startup, a window will appear to select the road zone (ROI).
Click on the corners of the road (minimum 3 points), then press ENTER.
Press ESC to skip ROI (the whole frame will be used).

python src/main.py photo.jpg

The image will be analyzed, results printed to the console, and the following files will be created:

• <image_name>_annotated.<ext> — image with annotated objects
• <image_name>_stats.json — detailed statistics

• traffic_log.csv — step‑by‑step statistics (timestamp, frame, vehicle/pedestrian counts, skipped objects, crossings, average speed, phase, incidents)
• dashboard_data.json — data for a web dashboard (latest snapshot and history)
• <image_name>_stats.json — photo analysis statistics

trafic-detector/
├── docs/                       # Documentation (optional)
├── src/
│   ├── dataset/                # Datasets folder (optional)
│   ├── console_dashboard.py    # Rich dashboard and session summary
│   ├── detection_utils.py      # Helper classes (incidents, eco, DB)
│   ├── main.py                 # Entry point
│   ├── photo_mode.py           # Photo processing
│   ├── requirements.txt        # Dependencies
│   ├── roi.py                  # Region of Interest selection
│   ├── traffic_light.py        # Traffic light class and timing calculation
│   ├── video_mode.py           # Main video/stream processing loop
│   └── visualization.py        # Drawing functions (HUD, boxes, graphs)
├── LICENSE                     # MIT License
└── README.md                   # This file

This project is licensed under the MIT License. See the LICENSE file for details.