Search
  • Alka Asthana

Image Recognition: In the eye of the beholder

Updated: Nov 9, 2020




Ever wondered how a machine is able to recognize an image. And of course, as always there is a Do Re Me! You can almost hear it humming “Let me start at the very beginninga very good place to start.”


The machine initially takes a few baby steps - and identifies the edges & contours of the image (in the early hidden layers of the neural network). Once it builds a simple representation – it delves deeper – just as we would. It takes this simple representation & then builds the complex representation (in the later hidden layers). Finally, it is able to identify the object.


Sounds deep, doesn’t it? You gave an input image & the machine gave you the output directly. Facebook uses this in their tagging algorithms.


The Seed Company by EW Gaze, a Canadian garden store, wanted to advertise its Walla Walla onion seeds on the social network, but its attempt was flagged as objectionable by Facebook’s algorithm.


We have long ways to go … but we are on the right path where we have moral, ethical and privacy considerations at the front & center of it all!


12 views0 comments

Recent Posts

See All