[Software] Microsoft launches Lobe, an app to train machine learning models with images

[Dark]

Microsoft has launched its new Lobe application, a free 'software' that allows anyone to train a machine learning model, for the development of Artificial Intelligence (AI), simply through the use of images.

   Lobe allows people "without any knowledge of data science" to import images and tag them to create a database that can be used to train a machine learning model, according to Microsoft, which acquired the developer of the app in 2018, in a release.

 

People can evaluate the pros and cons of the model with visual results in real time, as well as put them to the test and offer their opinion to improve the results of AI.

  The application automatically selects the most suitable machine learning architecture for each case and begins training the model without the user having to previously configure it or write code.

   After training is complete, machine learning models can be exported for use in applications, websites, and devices, so that both industry and users can create their own solutions.
   Among the tools that can be developed using Lobe, Microsoft has highlighted the possibility of creating alerts to warn that an employee is performing a dangerous task without a helmet or warn every time a wild animal approaches a garbage container, for example.

   As of this week, Lobe is available for free as a trial for Windows and Mac computers. At the moment it only has support for image classification, but Microsoft plans to expand it to other data types and models in the future.