Google has open-sourced its AI-powered tool, the secret behind Pixel’s so good portrait mode

The Pixel phone of Google has recently unveiled its much-touted camera powered by AI. Google is now ready to open-source its AI-powered tools that played a masterstroke for the success of Pixel’s portrait mode. The leading company has put forward it’s machine learning tool to capture the optimum out of its tiny smartphone lens.



Google is working on the machine learning process to increase its image segmentation capability. While this tool is created using convolutional neural networks(CNNs) which helps in enhancing the analysis of image segments aiding the software to split the image apart. It then blurs the background of the image leaving behind the subject pin-sharp.


The image segmentation tool performs an analysis of the visual data captured. It then separates the background elements from the foreground elements splitting them apart.

Earlier, Google has made available its AI-powered tool on the open-source which goes by the code name DeepLab-v3+. This tool based on machine learning process is built using CNNs. It was announced in a blog post earlier this week.


A highly enhanced feature carved out of the machine learning has made it cake-walk for the software to analyze visual data through a single lens. The shot taken by its portrait mode has unveiled high-end visual data for its user. Recently the shot was posted to help users understand their desires.


Meanwhile, the method may seem trivial but it adds a great skill to enhance the working of a camera. These fall into bokeh-style pictures which leave aside the subject pin-sharp making the background image blur.


And, Google has come forward to cash in the new tech available. While Apple iPhones popularized this feature worldwide followed by its distinct smartphone success.

Lately, Google seems to take a step ahead of Apple in this feat. As Google is creating the portrait effect on single lens compared By the double lens used by Apple.  


Google software engineers Liang-Chen and Yukun Zhu boasted of the new feat achieved due to the recent boom in machine learning. It was hard to expect something like this five years ago.

They feel optimistic about Google’s work to flow across industries with Google open-sourcing its new tool.