Fungi Classification using YOLOv8 with Auto-Labeling

This project demonstrates the use of YOLOv8 for classifying fungi images with the help of auto-labeling techniques using Label Studio. The methodology includes the setup of a labeling environment, training a deep learning model, and evaluating its performance.

Repository

The full guide for using Label Studio for auto-labeling can be found in the repository: Import-Annotated-Data-LabelStudio. This is a crucial resource for understanding how to automate the annotation process, which is essential for this project.

Project Overview

The goal of this project is to classify fungi images using YOLOv8, leveraging auto-labeling techniques to create a robust and efficient pipeline for data annotation and model training.

Methodology

1. Setup Label Studio for Auto-Labeling

• Create the necessary directory structure for datasets.
• Configure Label Studio to work with local files.
• Use a Python script to correct and manage label files.

2. Data Preparation

• Load the dataset and organize it into training and validation sets.
• Use Label Studio to annotate the images automatically.
• Validate and correct annotations as needed.

3. Model Training with YOLOv8

• Install the required libraries and dependencies.
• Train the YOLOv8 model on the annotated dataset.
• Monitor training progress using TensorBoard.

4. Model Evaluation

• Evaluate the model performance on the validation set.
• Visualize the results, including confusion matrices and sample predictions.

Evaluation Results

Conclusion

This project demonstrates the integration of auto-labeling techniques with deep learning models to classify fungi images efficiently. The use of YOLOv8 for training and Label Studio for annotation provides a powerful combination for developing robust image classification models.

References

• Import-Annotated-Data-LabelStudio