AlphaBit - OpenML

Training Structure

Directory Structure For Machine Learning Training

OpenML/ #Main folder for the OpenML project
					└── datasets/ #Folder containing training & validation datasets
						└── AI/
							├── train/
							│   ├── images/  #Images for ML training
							│   └── labels/  #Each image is labeled for ML
							└── val/
								├── images/  #Images in val are used for validation
								└── labels/  #Each image in val is labeled for ML

Detailed Explanations:

Main "OpenML" Folder: This is the root directory containing all necessary data and subfolders.

"datasets/AI": This directory contains the dataset used for training the AI model.

"train" and "val" Subfolders:

train: Contains the data the model uses to learn (e.g., images and labels).

val: Contains data used for model validation, i.e., to evaluate its performance on new data it hasn't seen during training.

In each "train" and "val" folder:

images: Here are the actual training/validation images.

labels: Here are the label files (annotation files) describing objects in the images.

Labels for YOLOv8n

To train a YOLOv8n model, the files in the labels folder must contain labels in the appropriate format:

Label Format (YOLO format):

Each line in a label file represents a detected object.

The format is:

<class_id> <x_center> <y_center> <width> <height>

Coordinate values (x_center, y_center, width, height) are normalized (divided by image dimensions).

<class_id> is an integer representing the object class.

Support -> Discord

Choose Language / Alege Limba