Dataset Generation (Coming Soon)

We will soon support dataset generation. This will utilize data scraping and the latest advancements in generative AI to generate datasets full of high quality data for fine-tuning models.

Dataset (Manual Creation)

This is your single spot for everything data. If you are looking to create a custom dataset, navigate to the Datasets service and create a new dataset. Inside Dataset you will notice a few aspects.

Object recognition

We are mainly focused on object recognition applications which involve bouding boxes around the specific objects in an image. Our datasets are currently oriented towards this and involved 3 main aspects

Entry

An Entry is a single data entry into your dataset. Currently, only image entries are supported. Inside the dataset press the upload button to upload images as entries. Then you are able to view and manage them in your dataset view

Class

Classes define the different objects you want the computer vision to be able to detect. It allows differentiation for training, so for example, if you want a model to recognize the different kinds of food, your classes would be the food names (Steak, Pasta, Apple, Etc)

Label

Labels define the location and type of an object in an image (eg. for food: Apple, Orange, Pasta). A corresponding bounding box given by normalized coordinates from the image, would be included with the object type. For simplification we have a built-in Annotator to simplify this labelling process!