Google has announced that the Google Cloud Video intelligence API is now in public beta and the Google Cloud Vision API has been updated with enhancements to existing features.
The Google Cloud Video intelligence API was first released in March 2017. The API uses machine learning to make videos searchable and easy to find by extracting metadata that can be used to index video content. The API annotates videos stored on Google Cloud Storage identifying key entities within the video and as they occur.
Extensive machine learning or computer vision knowledge is not required to use the API as it features a library of 20,000 labels. The API uses these labels to identify entities within the content of video. By extracting metadata from video content, the API can search video in a fashion similar to searching text documents. Developers integrate the Google Cloud Video intelligence API with applications to add capabilities such as content moderation, content recommendations, and contextual advertising. The beta release also includes pornographic content detection.
Google has updated its Cloud Vision API with enhancements to existing features. Enhancements include (but not limited to) increased label detection model vocabulary (now more than 10,000 entities), improved text detection model specifically regarding recall on Latin languages, and the addition of Google Image Search-powered web detection. The API now features web and document detection for general use which allows users to search for and find similar images on the web as well as the associated video metadata.
The updates to the Google Cloud Vision API will be rolled out within the next three weeks.