Cognitive Robotics Pen Detection

Aim

To build and evaluate an object detection system utilizing the YOLO (You Only Look Once) model for real-time object detection.

Components

• Dataset: Images captured using a Raspberry Pi camera.
• CNN Model: YOLO model for object detection.

Overview

This project leverages the YOLO model for detecting objects in images. The model is trained on a dataset created using a Raspberry Pi camera, capturing images of objects that are then annotated and processed for training.

Procedure

1. Image Capture: Capture images using a Raspberry Pi camera and the libcamera module.
2. Annotation and Dataset Preparation: Use Roboflow to annotate the images and prepare the dataset.
3. Model Training: Utilize the YOLO model to train on the prepared dataset.
4. Prediction: Run the model on test images to detect and visualize objects.

Running the Code

To execute the project, follow the steps outlined in the provided Jupyter notebook. You can also run the notebook in Google Colab for easy access and execution.

Results

The YOLO model successfully detects objects, and the results are visualized with bounding boxes on images. The following performance metrics were observed:

• Precision: 0.80
• Recall: 0.67
• Accuracy: 0.57

Confusion Matrix

	Predicted: No Object	Predicted: Object
Actual: No Object	4	15
Actual: Object	18	0

Example Outputs

Here are some examples of the model's predictions:

Conclusion

This project demonstrates the capabilities of the YOLO model for real-time object detection, showcasing its effectiveness in visualizing detected objects with bounding boxes.