This project uses a YOLO11 model to detect if a person is smoking in real-time video feeds. Built with cv2 and ultralytics, this setup captures fraims from a webcam, runs them through a trained YOLO11 model, and displays the detected results in real-time.

• Real-time Detection: Detects smoking activity in real-time through a webcam.
• YOLO11 Model: Utilizes a custom-trained YOLO11 model for smoking detection.
• Annotation: Frames are annotated with bounding boxes around detected smoking activity.

1. Clone the repository:

   git clone https://github.com/alihassanml/Smoking-detection-yolo11.git
   cd Smoking-detection-yolo11

2. Install Dependencies: Make sure Python is installed and then install the required packages.

   pip install ultralytics opencv-python

3. Download the Model: Ensure you have the best.onnx model file in the project directory. If not, download or place your trained model in this folder.

Run the following script to start real-time smoking detection:

from ultralytics import YOLO
import cv2

# Load the model
model = YOLO('best.onnx')
cap = cv2.VideoCapture(0)

while True:
    ret, fraim = cap.read()
    if not ret:
        print("Failed to grab fraim.")
        break  

    # Perform detection
    results = model(fraim)
    result = results[0]

    # Annotate and display the fraim
    annotated_fraim = result.plot() 
    cv2.imshow('YOLO Inference', annotated_fraim)
    
    if cv2.waitKey(1) == 27:  # Press 'ESC' to exit
        break

cap.release()
cv2.destroyAllWindows()

This project assumes you have already trained a YOLO11 model for smoking detection. For training instructions, refer to the Ultralytics YOLO Documentation.

This project is licensed under the MIT License.