Twitter brings smart auto-cropping for your photo posts with the help of machine learning

Our daily social media photo uploads do not necessary contain people in them all the time. Sometimes it’s just a scenic view, while other times there are our furry fellows too. [caption id=“attachment_3864989” align=“alignleft” width=“380”]  Representative Image[/caption] Keeping this in mind, Twitter has come up with a smart auto-cropping features that uses the power of machine learning to frame the right bits of shared photographs. So your followers at a glance can see what makes the image, instead of a random background. According to Twitter, the system will crop images with the help of the neural network. Here, it uses a picture “saliency quality” which can help in determining attributes of the picture, ones that would normally attract attention. These attributes would be included in the pictures and cropped accordingly. Twitter has used data from eye-tracking studies which has helped them to form various predictions for smart cropping. They found that puppies, faces, text, and regions of high contrast attracted the most attention from people. With the help of this data, the system would predict which part of the image would be cropped. [caption id=“attachment_4320667” align=“alignnone” width=“825”]  Twitter uses machine learning to autocrop the images. Twitter.[/caption] Dubbed ‘smart auto-cropping’, this can be helpful for those images where there may not be facial images and could be scenic view or an animal. This feature has been rolled out in Twitter’s app on iOS and Android. For now, it is limited only to pictures and not videos. Meanwhile, Twitter is expected to come with a new feature where users would be able to **shoot videos from the app itself** , just like Snapchat.