Artificial Intelligence requires high quality data to deliver accurate results.

AI is intelligent indeed, but it is also dumb in some ways.

One cannot just rely on artificial intelligence to do everything on its own, it needs certain inputs of high quality that can allow it to work smoothly and deliver the expected results.

Why Facial Recognition Technology works so well with the iPhone Face ID

Facial Recognition technology is a phenomenal example of artificial intelligence but only if the pictures it reads are of high quality. This is why it cannot properly distinguish a thousand faces if the picture quality is low.

You must be wondering now why does iPhone Face ID work so well?

Well, it only works so well because the camera is extremely near the face of the user. This allows the whole system to take a high-quality picture and store an extremely accurate mathematical representation of the face. iPhone has an infrared camera too. A normal camera with an infrared camera placed extremely near the face deliver extremely amazing results. Both of these features are combined to ensure high quality of facial recognition, therefore, making Face IDs an acceptable part of the security of the phone.

How does Facebook photo tagging deliver accurate results?

Facebook photo tagging works extremely well. It delivers rediciously good results. It works so well that one thinks artificial intelligence can recognise everyone’s faces. However this is not true.

You may have 2000 friends on your friends’ list, but there must be a certain number of people that you regularly interact with, let’s say that is 20. This makes Facebook’s life easy; it can just limit the search to few people and then search for the faces in a photo, in that narrow list of 150 people. So instead of looking into those 2000 faces of all your friends, Facebook makes its life super easy and only searches 20 faces and delivers almost accurate photo tagging.

Poor Quality data is a big no for facial recognition technology

Facial technology is smart but it’s difficult to use sometimes when data is of poor quality, the current algorithm needs high-quality data. It needs high-quality image data and also high-quality data to recognise once it has been trained.

This is why street cameras are not likely to recognise faces on their own.

Street cameras are just not good enough to capture high-quality videos. The current algorithm just cannot distinguish between faces that are similar to each other when viewed from far.

This is why artificial intelligence is not able to deliver good results when the images are captured from street cameras. Street cameras record an extremely low quality footage. Artificial intelligence does not deliver great results when used with poor quality footage.

If you need help with understanding whether you need to use Artificial Intelligence and if you do need to use it, to what extent, we can help. Book your free session now!

Zain Daniyal

Data Consultant