Mindtech Global, developer of the world’s leading platform for the creation of synthetic data for training AI, has released a significant update to its Chameleon platform. The end-to-end platform now enables the automated creation of millions of individual ‘actors’, placed into virtual worlds, creating synthetic data for training AI visual systems addressing a wide range of markets, including retail, security, smart city and home. This news follows the recent announcement of $3.7 million investment round led by strategic partner Appen, the global leader in data for the AI Lifecycle.
Using Mindtech’s Chameleon platform to train AI systems lessens developers’ dependance on real-world data – which is often scarce, time consuming to annotate and requires strict compliance to ever-changing data privacy regulations. Chameleon’s ‘no-code’, ‘self-serve’ platform allows AI developers and data scientists to quickly automate the creation of high-quality, precisely annotated and privacy-compliant synthetic data by building computer generated scenes and scenarios. Mindtech’s unique, behavioral-led, simulation platform leads to faster creation of relevant data, promoting an “intelligently engineered data” approach to training vision systems.
This latest platform release sees the introduction of configurable “actors” – automated photo-realistic avatars, who can act and interact within a given scene and scenario – and offers users a wide range of diversity options including clothing, hairstyles, glasses, face coverings, height, build, age, and skin tone. The synthetic data generated enables AI vision systems to recognise greater diversity, visualise crowds and detect individual actions more accurately.
View a video of actor configurations available on the Chameleon platform: https://vimeo.com/showcase/mindtechchameleon221
Chris Longstaff, Mindtech’s VP Product Management said, “This new Chameleon release was driven by customer demand for greater diversity in AI training datasets, to ensure robust solutions that better address corner cases and bias. The ability to rapidly create a dataset with required human diversity statistics, such as age, race, fashion and regional characteristics, is essential for creating bias-free AI systems that better understand how humans interact with each other and the world around them.”
Chameleon 22.1 is available for immediate licensing.
Mindtech is exhibiting at NexTech Week AI Expo Tokyo on stand 7-18 between 11th and 13th May.
