Alegion launched its new suite of image and video annotation tools for training data in computer vision initiatives. These new capabilities are specialized for data tasks like image classification, object localization, and semantic segmentation, and are being used by customers across retail, automotive, technology, government, and financial services.
Alegion’s enterprise computer vision suite features machine learning-assisted capabilities like object proposal in imagery, object tracking in video, and superpixel algorithms for semantic segmentation that increase efficiency and accuracy in image and video annotations. These capabilities are integrated with other Enterprise platform features like iterative task workflows, complex taxonomy support, nested entity relationship classifications, and machine learning-augmented quality controls, all designed for the most sophisticated computer vision data annotation use cases.
“Enterprise-class use cases in computer vision require advanced classification capabilities and feature images and videos that are target-rich, necessitating sophisticated workflows that integrate the best of human annotations with machine learning to efficiently produce accurate results,” said Chip Ray, CTO of Alegion. “The combination of machine-assisted annotation, iterative workflows, and machine learning-augmented quality controls in the Alegion platform have proven to be effective at delivering accuracy at scale.”
Numerous customers are already experiencing success with new computer vision capabilities from the platform. “High-quality training data is key to the development of our artificial intelligence model. Alegion’s image annotation solution has accelerated our training data preparation and efficiently produces data that meets or exceeds our accuracy requirements. Alegion’s annotation tools, machine learning assistance, and quality control strategies have greatly improved confidence in our AI initiatives,” said Steve Muscarello, Vice President of Product Development at Insurance Auto Auctions, Inc.