Google’s Pixel phone is one of the best phones that is available in the market. The Pixel phone got a camera that is best of its kind present in the market, and this is possible due to AI. Google has used Al tools to take better shots and shooting modes through a small smartphone camera lens.
As per a blog, Google announced that the Al software which is used for Pixel’s portrait mode has been open-sourced. The code which is open-sourced by Google is named as DeepLab-v3+. This is a tool used for image segmentation and is built by using convolution neural networks (CNN). CNN is a machine learning language that is good at analyzing visual data. The work of the image segmentation is to analyze the object in a picture and split them to foreground elements from background elements.
This process is very useful for the cameras and Google is using it in its Pixel’s camera portrait mode. This type of mode gives a bokeh-style image in which it blurs the background and makes the subject pin-sharp. This effect was bought by Apple iPhone which uses two camera lenses to have a portrait effect. The Google Pixel camera uses just one lens to give us this portrait effect.
Google software engineers Liang-Chieh Chen and Yukun Zhu said the quality of the image segmentation has improved to such a level with a deep-learning boom, that the accuracy level we have now is hard to get before five years. It is said by the company that, it is sharing the data publicly so that other researchers and industry can avail this data to improve the quality further on the work of Google. The open-source of software to the public will help the developers to make their own image segmentation.