Sama, the leader in providing data annotation solutions that power the AI models of the future, announced Platform 2.0, a re-engineered computer vision platform to reduce the risk of machine learning (ML) algorithm failures. The completely redesigned, scalable platform, consisting of SamaIQ™, SamaAssure and SamaHub, offers greater transparency for clients to minimize rework and allows Sama to deliver annotated data and insights three times faster. Platform 2.0 has successfully achieved a 99% client acceptance rate for AI training data through SamaAssure, the industry’s highest quality guarantee, with an annotation delivery rate of up to 300+ million frames, 850+ million shapes and 10 billion annotation points a month. With these enhancements, Sama clients can now deploy ML faster than ever, with the assurance that their AI models are trained on exceptional-quality data to avoid concerns of drift and/or bias.
“As companies continue to use machine learning algorithms to power some of the most innovative technologies today, our collective responsibility as leaders in tech is to ensure the applications are built using high-quality, unbiased data,” said Wendy Gonzalez, CEO of Sama. “Without accurate inputs, AI applications won’t be effective and may even yield potentially dangerous outcomes. Our Platform 2.0’s value lies in both its exceptional data accuracy and its unique combination of approach, people and technology.”
Also Read: Axonify Evolves Artificial Intelligence Capabilities, Launching New Generative AI Tool
According to McKinsey, just 36% of ML algorithms successfully transition beyond the pilot stage, indicating that more than 60% of ML algorithms do not deliver the intended business impact. Sama helps companies reduce the risk of ML model failure through a unique combination of cutting-edge technology, actionable data insights delivered by SamaIQ and human expertise. Platform 2.0 can quickly classify similar images based on a number of factors, surfacing unique images to the Sama team for faster identification and recognizing if a dataset has too many similar images. SamaIQ further speeds up the process by proactively identifying actionable insights to get models to production, including early edge case identification, data trends and analysis, missing data identification and feedback on model performance issues. Finally, Sama proactively calibrates the platform to a client’s specific needs and leverages a proprietary Human in the Loop (HITL) approach that includes specialized training for each project annotators work on.
“Sama helped us build an incredibly solid data set to make our immersive seats and haptic platforms more responsive for motion in theaters and simulations. They quickly annotated training data for our ML models with consistency and accuracy. Their team of annotators was very well trained, adapting quickly to our specific needs, even the more challenging ones. Overall, we were very impressed with the quality control and volume of work completed,” said André Beaudin, Engineering Manager, AI & Advanced Processing at D-Box.
SOURCE: Businesswire