Innovation That Matters

Lobe's machine learning interface

Machine Learning platform is easy as drag-and-drop for developers

Computing & Tech

A startup has developed a cloud-based, machine learning SaaS platform that enables users to build their own models without having to code.

Machine learning is a tool many industries are adopting. The software uses AI (Artificial Intelligence) to analyze datasets, learning as it does so. It is a rapid and accurate tool used in fields as wide as medicine and construction.

Lobe is enabling businesses to rapidly integrate machine learning technology into their apps and platforms. Lobe’s software doesn’t require users to code anything. Instead, it uses a simple drag-and-drop interface. Users begin by dragging in a set of images. This is a ‘training’ dataset. Lobe then generates a deep learning model, which can perform image recognition on new, unseen datasets. Users can look into any aspects of their data with easy to use visuals.

Users can control their own model sensitivity levels and also the number of fields the model detects. For instance, the model can be trained to recognize both hands and facial aspects, too. Model training and analysis takes place in a cloud platform. This prevents users’ hardware from being slowed down by the calculation each model completes. The cloud will also store data remotely, for added security.

Lobe has made exporting datasets straightforward, too. Users can export their models into Tensorflow (which Lobe uses as the architecture for its models). Alternatively, the Lobe developer API enables users to transfer models into their software as they please.

Lobe has developed a range of image-recognition tools to show what’s possible. For example, one suite can recognize hand gestures and translate them into corresponding emojis. Another can gauge the weight of a pile of ground coffee.




Download PDF