Google’s March 8, 2017 machine-learning announcements did not put every product at the same stage. Cloud Video Intelligence API was in private beta, while Cloud Machine Learning Engine was generally available. A separate post announced that Kaggle would join Google Cloud. A developer’s reference sheet needs three rows, because each announcement supports a different next step. [video][kaggle]

The video row concerns an existing-model service for finding content. Google described locating shots containing entities across videos stored in Google Cloud Storage and identified media catalogs as a use case. Label that row private beta: a developer would need to establish access before testing it. The retained announcement does not specify an application procedure, and the capability description is not an independently measured accuracy result. [video]

The engine row concerns training and deploying a developer’s own TensorFlow models through a managed cloud service. Google marked Cloud Machine Learning Engine generally available. That is the deployable service in this comparison, rather than evidence that the video API had left beta. The distinction also concerns the work: custom-model development and calling a supplied video-indexing capability are different routes. [video]

The Kaggle row belongs under announced plans. Google said Kaggle would join its cloud business and described continued training and deployment services together with community dataset storage and querying. Track implementation details rather than treating those commitments as proof of a completed integration. Together, the rows separate access to investigate, a generally available development service and an organizational announcement. Both sources are Google’s account, not independent confirmations. [kaggle][video]

Source documents

  1. Google Cloudhttps://cloud.google.com/blog/products/gcp/announcing-google-cloud-video-intelligence-api-and-more-cloud-machine-learning-updates
  2. Google Cloudhttps://cloudplatform.googleblog.com/2017/03/welcome-Kaggle-to-Google-Cloud.html