YOLOv7 Workflow: Dataset Creation, Training, and Deployment

Dataset preparation

Let’s look at how we can create a dataset using open images. In the openimages/src/prepare_dataset.ipynb file, we will download a dataset of three classes: airplane, bus, and bicycle. The dataset also contains their object detection annotations in YOLO format. As the next step in the same tutorial, we will discover how to prepare the extracted dataset into the correct structure to use directly for the training step.