This project demonstrates how to use MediaPipe's Object Detection model along with OpenCV for real-time object detection on images. The code loads an object detection model, processes an input image, and visualizes the detected objects by drawing bounding boxes and category labels.

Before running the code, ensure you have the following installed:

1. Python 3.7 or later

2. Dependencies: Install the required libraries using pip:

   pip install mediapipe opencv-python-headless numpy

3. MediaPipe Model: Download the required model.tflite file for object detection and place it in the project directory. Update the model_asset_path in the code if necessary.

4. Image File: Ensure you have an image at ./train/images/IMG_0509.jpg or update the file path in the script.

1. Import Libraries:

   • mediapipe for object detection.
   • opencv-python for image processing.
   • numpy for array manipulations.

2. Object Detector Initialization: The MediaPipe Object Detector is initialized with the model.tflite file and configured to process images.

3. Image Preprocessing:

   • Load the image using OpenCV.
   • Convert it from BGR to RGB format for compatibility with MediaPipe.
   • Ensure the image is in the correct format (uint8 with shape (H, W, 3)).

4. Object Detection: The image is passed to the detector, and the results are processed to draw bounding boxes and labels on the image.

5. Results Visualization: The processed image with detections is displayed using OpenCV.

1. Clone or download this repository.
2. Place your object detection model (model.tflite) in the root directory.
3. Place your test image in the ./train/images/ directory or update the path in the script.
4. Run the script:

   python object_detection.py

5. View the output image in a pop-up window. Press any key to close it.

Detected objects will be highlighted with green bounding boxes, and their categories along with confidence scores will be displayed on the image.

• If the image fails to load, a ValueError is raised.
• The script includes checks for valid image format and dtype. If these are not met, appropriate error messages are shown.
• Errors during MediaPipe image creation or object detection are handled and displayed.

• Python
• MediaPipe
• OpenCV
• NumPy

To install the dependencies, run:

pip install mediapipe opencv-python-headless numpy

1. Change Input Image: Replace the file path in cv2.imread() to point to a different image.
2. Adjust Model: Replace model.tflite with a different MediaPipe model if needed.
3. Modify Confidence Threshold: Change the if confidence >= 0.50 line to adjust the detection threshold.
4. Bounding Box Color and Style: Edit the cv2.rectangle and cv2.putText parameters to customize the visualization.

This project is open-source and free to use for personal and educational purposes. Check the respective licenses for MediaPipe and OpenCV if using in commercial projects.