Rise of the Robots

05 January 2018

SIG: Contemporary

The magic wand symbol in Google’s Snapseed or Adobe Lightroom CC both automatically adjust the tonality of an image.  Artificial intelligence, also known as “machine learning”, is likely to be behind their image enhancement tools.  Adobe say that their Sensei tool (built-in to Lightroom CC) re-creates elements in photos where they didn’t exist by studying nearby pixels.

The Sensei tool uses the power of machine learning to identify objects in images and automatically adds searchable words to photo tags, plus it recognizes faces so you can find all your pictures of a specific person quickly. 

I use both Snapseed (on my mobile devices) and Lightroom (on both my mobile and desktop computers) and welcome the help that the auto-fix tools provide in getting raw images ready for fine adjustments to my taste.  I’ve also found the ability of Lightroom to “find all cats” in my library of pictures useful when creating collections of themed images.  (Ok, it’s not just for cats, our group could use it to find chairs.)

Google recently announced a project “NIMA: Neural Image Assessment” which has been trained in assessing both image technical quality and subjective image aesthetics.  The NIMA tool scores images based on an Aesthetic Visual Analysis (AVA) algorithm developed by researchers at the Computer Vision Center of the Universitat Autonoma de Barcelona (Spain) and Xerox Research Centre Europe (France).  Tests apparently show NIMA scores images similarly to a human using the same criteria.  The Google Research Blog mentions NIMA’s future use in intelligent photo editing tools. 

One fear I and many others may have is this: will automation with artificial intelligence in our editing tools lead to uniformity in photography?  In some on-line forums I’ve seen people asking for NIMA capability built-in to their cameras… Are we witnessing the rise of the robot and the death of originality?

I wonder how the NIMA algorithm would score Alec Soth’s images in Sleeping by the Mississippi?

Further reading below:

Project NIMA




Google neural network able to identify location of images


Adobe Sensei



Image from Google Research Blog, containing images from on-line communities used for AVA algorithm development.