YOLOv8-Dataset-Transformer is an integrated solution for transforming image classification datasets into object detection datasets, followed by training with the state-of-the-art YOLOv8 model. This toolkit simplifies the process of dataset augmentation, preparation, and model training, offering a streamlined path for custom object detection projects.

• Dataset Conversion: Converts standard image classification datasets into YOLOv8 compatible object detection datasets.
• Image Augmentation: Applies a variety of augmentations to enrich the dataset, improving model robustness.
• Model Training and Validation: Facilitates the training and validation of YOLOv8 models with custom datasets.
• Model Exporting: Supports exporting trained models to different formats like ONNX for easy deployment.

• Python 3.8 or later
• PyTorch 1.8 or later
• YOLOv8 dependencies (refer to YOLOv8 documentation)

Clone the repository to your local machine:

git clone https://github.com/[YourUsername]/YOLOv8-Dataset-Transformer.git
cd YOLOv8-Dataset-Transformer

Install the required packages:

pip install -r requirements.txt

1. Prepare Your Dataset: Place your image classification dataset in the designated folders.

2. Run the Dataset Preparation Script:

   python dataset_preparation.py --markers train20X20 --irrelevant irrelevant --output output --total_images 1000 --train_ratio 0.8

   

   New images shall be generated by the script, you can refer to the image above:

3. Train Your Model:

   python train.py --data_config path/to/data.yaml --epochs 100 --model_name yolov8n.pt

4. Evaluate and Export Your Model:

   Validate, predict, and export your model using options in the train.py script.

Train and demonstrate the model and computed the parameters of experiments.

Contributions to the YOLOv8-Dataset-Transformer are welcome! Please read our Contributing Guidelines for more information. We fetch the train20X20 dataset from apoorva-dave and irrelevant images from Google image.

This project is licensed under the Apache 2.0 License - see the LICENSE file for details.

• Thanks to the Ultralytics team for the YOLOv8 model.
• Special thanks to all contributors and maintainers of this project.