Google’s Cloud AutoML simplifies artificial intelligence; makes machine learning easier to implement!
Machine learning is an interesting field, as it offers the opportunity to transfer the burden of labor-intensive tasks to a computer, reducing the costs and man-hours needed. While it may sound great, the actual process of implementation is much more difficult, with the costs involved with hiring engineers to develop your machine learning model. With Cloud AutoML, you can now easily customize your machine learning model with an easy to use GUI.
The first product of Cloud AutoML that Google will be introducing will be the Cloud AutoML Vision. This services allows you to create custom ML models for image recognition. Upload images, train and manage your models, and deploy those trained images directly on Google Cloud. Early tests have shown that Cloud AutoML Vision is more accurate at classifying public datasets than generic ML APIs.
The Cloud AutoML Vision is built on Google’s image recognition technologies, including transfer learning and neural architecture search technologies. This will allow you to easily create an accurate model even with limited machine learning expertise. The simple GUI also allows you to create a quick model in minutes, or create a full, production ready model in just a day.
Urban Outfitters has implemented Cloud AutoML to automate the product attribution process by recognizing product characteristics like patterns and neckline styles. This allows them to cut out manually creating product attributes, which are necessary for a great recommendation experience. Meanwhile, the Zoological Society of London has implemented Google CloudML to tag the millions of images captured by camera traps in the wild.
AutoML Vision is but the first of an array of Cloud AutoML products in development. If you want to try out AutoML Vision for yourself, you can request access via this form.
Pokdepinion: AI vision is a cool feature to have, and certain smartphones like the HUAWEI Mate 10 already has it. The full potential of AI and ML remains to be explored, so it’s nice to see Google making it easier for business to implement it.