How to train and classify images using Google Cloud Machine Learning and Cloud Dataflow Google Cloud Big Data and Machine Learning Blog Google Cloud Platform

For specialized image-classification use cases, using Google Cloud Dataflow and Google Cloud Machine Learning makes it easy to train and implement machine-learning models. Google Cloud Vision API is a popular service that allows users to classify images into categories, appropriate for multiple common use cases across several industries. For those users whose category requirements map to the pre-built, pre-trained machine-learning model reflected in the API, this approach is ideal. However, other users have more specialized requirements -- for example, to identify specific products and soft goods in mobile-phone photos, or to detect nuanced differences between particular animal species in wildlife photography. For them, it can be more efficient to train and serve a new image model using Google Cloud Machine Learning (Cloud ML), the managed service for building and running machine-learning models at scale using the open source TensorFlow deep-learning framework.