Datasets are the core of every deep learning project. They often contain tens of thousands of images spread through tens of teams for annotation and quality control. Dataset management helps you with integrations with other services as well as performing time consuming tasks in the matter of seconds.

Data augmentation tools

Dataset Versioning

Versioning to keep records of changes and their impact

Merge & duplicate

If you need to combine two datasets, all you need is a single click

Import / Export

Import or export from/to most of the standard formats easily

Hassle free data management

When dealing with large amounts of data, stretched over different teams, it’s really easy to get lost. That’s why we implemented data versioning and data management tools to help you with it. Not only our tools help you easily perform time consuming tasks in matter of seconds, but it will also help you keep track of every single change and how it affects your results.

Keep different versions for each project’s dataset. Perform trains and compare results to see the impact of incremental changes on the results.

Versioning & comparison

How big organizations or giant projects operate? By dividing the problem and conquering it. Our versioning system works much like that. Just like software engineering, you can build different versions of your dataset with differences in parameters of your choice to help you with managing your goals and keeping track of what you do and their impact on the results.
Our version comparison tool beaks down the differences between two different versions and compare the train results of each version to give you a better idea of the impact of the differences on the results.

Merge & duplicate

What seems like an easy thing to do actually has some tricks up its sleeve. Like what happens to all the tasks team assignmentsm or what happens if you want to merge an object detection dataset with a segmentation dataset. More challenging is when you want to merge two overlapping datasets. Normally you have to painstakingly remove duplicate files nd then merge the datasets but we have implemented all the necessary tools to take care of all of these hassles and give you a smooth experience, undertaking all these operations in seconds with just a few clicks.

Dataset conversion

At times a dataset can be used for multiple model types. For example you can use the same model that detects car numbers to detect the presence of cars. One of the models is classification but the other is segmentation. Normally, you have to re-label all the images to be able to use them for the new model, or write script to convert them but we give you a powerful conversion tool that helps you convert any type of dataset to any other on the fly without losing your precious data.