This project detects and tracks vehicles in a video using YOLOv8 and adds an intelligent bridge control system:

• It continuously estimates the current load on a bridge.
• The bridge has a capacity limit of 20 tons.
• When admitting a new vehicle would exceed the capacity, the system blocks entry (no new vehicles can pass) until enough weight exits.

The video feed shows real-time status overlays: current load, capacity, and whether the bridge is OPEN or CLOSED.

• Vehicle detection (cars, buses, trucks) with YOLOv8
• Lightweight multi-object tracking to assign stable IDs per vehicle
• Entry/exit line logic to know when a vehicle enters/leaves the bridge
• Per-class weight estimates (customizable)
• Real-time bridge capacity management (default 20 tons)

1. Install Python 3.9+ (recommended)

2. Create and activate a virtual environment (optional but recommended)

python3 -m venv .venv source .venv/bin/activate

3. Install dependencies

pip install ultralytics opencv-python pandas cvzone

4. Run the app

python yolov8-multiple-vehicle-detection/mainh.py

You should see a window with the processed video. Press Esc to quit.

1. Read a video frame.
2. Detect objects using YOLOv8.
3. Keep only vehicles of interest: car, bus, truck.
4. Track each vehicle across frames to give it a stable ID.
5. Use two horizontal lines:
   • Entry line (vehicle enters bridge when crossing this line)
   • Exit line (vehicle leaves bridge when crossing this line)
6. When a tracked vehicle crosses the entry line:
   • Check its estimated weight (based on class) and the current total on-bridge load.
   • If admitting it keeps the total load <= 20 tons, it is allowed and the load increases.
   • Otherwise, the bridge is CLOSED for new entries (the vehicle is visually blocked) until weight drops.
7. When a tracked vehicle crosses the exit line and it was on the bridge, we subtract its weight.
8. The overlay shows:
   • OPEN/CLOSED
   • Current load vs capacity (tons)
   • A red barrier strip when CLOSED

flowchart TD A[Start / Read frame] --> B[YOLO detect vehicles] B --> C[Group by class: car/bus/truck] C --> D[Track each class -> stable IDs] D --> E[Compute vehicle center Y] E --> F{Crossed Entry line?} F -- Yes --> G[Would load + vehicle weight <= capacity?] G -- Yes --> H[Allow entry: add ID, increase load] G -- No --> I[Block entry: do NOT add] E --> J{Crossed Exit line?} J -- Yes --> K[If on-bridge: remove ID, decrease load] H --> L[Update overlay] I --> L K --> L L --> M[Draw status OPEN/CLOSED] M --> N[Display frame] N --> A 

Open yolov8-multiple-vehicle-detection/mainh.py and look at the top constants:

CAPACITY_TONS = 20.0 VEHICLE_WEIGHTS_TONS = { 'car': 1.5, 'bus': 12.0, 'truck': 15.0, } ENTRY_LINE_Y = 184 # Entry line position (in pixels, Y axis) EXIT_LINE_Y = 209 # Exit line position (in pixels, Y axis) LINE_TOLERANCE = 8 # Crossing tolerance (pixels)

• Change capacity: set CAPACITY_TONS to your desired limit.
• Adjust weights: edit VEHICLE_WEIGHTS_TONS to better match your environment.
• Move lines: tweak ENTRY_LINE_Y and EXIT_LINE_Y to match your scene.

Tip: Use the mouse to inspect coordinates in the window if needed (debugging hook exists in code).

• yolov8-multiple-vehicle-detection/mainh.py: main app with detection, tracking, and bridge logic
• yolov8-multiple-vehicle-detection/tracker.py: simple multi-object tracker used to keep stable IDs
• yolov8-multiple-vehicle-detection/coco.txt: class labels used by the YOLO model
• yolov8-multiple-vehicle-detection/tf.mp4: sample input video (you can replace it)
• yolov8-multiple-vehicle-detection/yolov8s.pt: YOLOv8 weights file (pretrained)

• A vehicle is considered “on the bridge” only after it crosses the entry line and is admitted.
• If admitting a vehicle would exceed the capacity, the system:
  • Keeps the bridge CLOSED to new entries
  • Shows a red barrier at the entry line and a message: “NO ENTRY - OVER CAPACITY”
  • Continues to allow vehicles already on the bridge to exit; once enough weight exits, the bridge reopens automatically.

This is purely a logic/visual block in this demo; it does not physically stop vehicles. In a real system, you’d connect this signal to a hardware barrier.

• Put your own video file in yolov8-multiple-vehicle-detection/ and update this line in mainh.py:

cap = cv2.VideoCapture('tf.mp4') # change to your filename

For live cameras, pass an index (e.g., 0) or RTSP URL to VideoCapture.

• Python 3.9+
• Packages:
  • ultralytics (YOLOv8)
  • opencv-python
  • pandas
  • cvzone

Install via:

pip install ultralytics opencv-python pandas cvzone

If you have a compatible GPU and want acceleration, follow Ultralytics’ docs for CUDA/torch install instructions, then install ultralytics.

• "No module named cv2" → pip install opencv-python
• "No module named ultralytics" → pip install ultralytics
• Video window not showing → some environments (like headless servers) can’t open GUI windows. Run locally or set up a virtual display.
• Model loads slow on first run → weights are large; after the first load it’s faster.
• Detections look off → ensure the video resolution fits the scene and lines (ENTRY_LINE_Y, EXIT_LINE_Y) are placed correctly.

• Added real-time bridge capacity logic with a default 20-ton limit.
• Added per-class weight estimates and gating logic at the entry line.
• Added OPEN/CLOSED overlay and a red visual barrier when closed.
• Switched to ID-based entry/exit accounting so load is accurate as vehicles leave.

This repository builds on YOLOv8 (Ultralytics). Review and respect their license when using the model weights and code.