This repo gives you the ability to connect a RoboFlow dataset to Kaggle notebok and using the Gpu's for training and evaluation

First let's install needed packages

!pip install roboflow -q
!pip install ultralytics -q

Grab the dataset link from roboflow

from roboflow import Roboflow
rf = Roboflow(api_key="your roboflow api key here")
project = rf.workspace("workspace").project("project name") 
version = project.version(4)
dataset = version.download("yolov9")

in order to change the yaml file configuration to match your data train test valid folders you will execute the following code

import yaml

def update_yaml_content(file_path, new_content):
    try:
        # Write the new content to the YAML file
        with open(file_path, 'w') as file:
            yaml.dump(new_content, file)

        print("File updated successfully.")
    except Exception as e:
        print(f"An error occurred: {e}")

# Define the yaml file path
file_path = '/kaggle/working/dataset/data.yaml'

# Define the new content
new_content = {
    'names': [
        'Early Blight',
        'Healthy',
        'Late Blight',
        'Leaf Miner',
        'Leaf Mold',
        'Mosaic Virus',
        'Septoria',
        'Spider Mites',
        'Yellow Leaf Curl Virus'
    ],
    'nc': 9,
    'roboflow': {
        'license': 'CC BY 4.0',
        'project': 'proj name',
        'url': 'url',
        'version': 4,
        'workspace': 'workspace'
    },
    'test': '/kaggle/working/dataset/test',
    'train': '/kaggle/working/dataset/train',
    'val': '/kaggle/working/dataset/valid'
}

# Call the function to update the YAML file
update_yaml_content(file_path, new_content)

File updated successfully.

check if yaml file is updated correctly or not

import yaml

def view_yaml_file(file_path):
    with open(file_path, 'r') as file:
        yaml_content = yaml.safe_load(file)
        return yaml_content

file_path = "/kaggle/working/dataset/data.yaml"
yaml_content = view_yaml_file(file_path)
print(yaml_content)

Train code for Yolo-V9

from ultralytics import YOLO

# Build a YOLOv9c model from scratch
model = YOLO('yolov9e.yaml')

# Build a YOLOv9c model from pretrained weight
model = YOLO('yolov9e.pt')

# Display model information (optional)
model.info()
# Train the model
results = model.train(data='/kaggle/working/dataset/data.yaml', epochs=300,
                    batch=32,patience=10,cache=True,save_period=30,imgsz=300)