Google has announced a number of updates to its Vision AI portfolio. Specific updates include enhancements to AutoML Vision Edge, AutoML Video, and the Video Intelligence API. All three machine learning products continue to grow in feature sets and the ability to readily impact the use cases where they are implemented.
The AutoML Vision Edge can now detect objects. Edge devices such as connected sensors and cameras can now utilize AutoML Vision Edge to detect changes, anomalies, failures, and other trigger events based on object detection. AutoML Vision Edge can both classify images and detect certain objects directly from the edge device (no need to rely on core technology for decision making). Hardware currently supported includes those that use NVIDIA, ARM, Android, iOS, and some other chipsets.
AutoML Video Intelligence can also detect objects now. Users can leverage this functionality to train models to create labels for certain content within a video. This allows users to track the movement of labeled objects from frame to frame. Specific use cases include traffic management, sports analytics, robotic navigation and other scenarios where objects move beyond the scope of a single frame.
The Video Intelligence API can now track and recognize logos. Out of the box, it recognizes over 100,000 popular business and organization logos. Logo identification adds to the existing recognition of objects, scenes, and actions. To learn more about any of these new features, visit the Vision website.