This week Mindtech announced the latest release of its Dolphin platform, which performs training data analysis, visualisation and curation of image datasets for AI-based computer vision.

The new Dolphin release enables data scientists and machine learning engineers to get an enhanced understanding of their training datasets. Engineers can rapidly analyse dataset shortcomings such as sparsity of data, data diversity issues and data-drift, when original training data is no longer relevant. In addition, visualisation of data annotations enables engineers to quickly review chosen images, such as outliers, and examine and verify the associated annotations.

Dolphin is part of the Mindtech DataOps platform, and the actionable intelligence created by Dolphin can be used in Mindtech’s Synthetic Data Creation Platform, Chameleon, to enable simple and rapid creation of synthetic data to improve training data coverage and AI model performance.

Mindtech attended CVPR Expo in Seattle on June 19th – 21st where the team demonstrated the new platform.

Read more here.