The most relevant options for facial recognition

It is possible through a photo, to know your mood, age, sex, context around you … the only doubt you may have is, what tool to use to carry it out?

A few years ago there was a stir with everything related to “Cognitive Services”, with certain features that allowed us to be closer to tasks carried out by the human brain. One of them was to be able to identify through images and videos, as a normal person does. 

At this time, we may already find the recognition cameras of people in airports, and even in some cases, the implementation that China is doing. But what can this type of algorithm recognize through an image?

• Age
• Emotions
• Environment
• Sex
• Know which person you are specifically
• Extra features added

In the beginning, these tools offered by large companies, had certain limitations and errors, and only returned certain characteristics. Over the years, we are able to detect more characteristics of the images without having to develop more functionalities.

This type of applications, we could divide them into two parts:

• The consumer of the API, which would be our web, mobile or server application, that sends the image somewhere and receives a response, in most cases .json
• The one that analyzes: the part that receives the image, processes it and returns a result

In most cases, we will develop the application that will consume these services and process the data, and we will only choose a provider of such technology and configure it.

Among the alternatives we have, and depending on what we want to analyze in the image, the budget we have, the importance of our data … we could choose between these options:

• Face API (Open Source): https://github.com/justadudewhohacks/face-api.js?files=1
• Open Face (Open Source): https://cmusatyalab.github.io/openface/
• Microsoft: https://azure.microsoft.com/en-us/services/cognitive-services/face/
• IBM Watson: https://www.ibm.com/watson/services/visual-recognition/
• Google Cloud: https://cloud.google.com/vision/

The payment tools have a free version up to a certain number of calls, in case you need to try some of its features.

Although this year the Blockchain and virtual assistants are booming with the themes of voice analysis, we do not have to forget that the eyes will always be the mirror of the soul, and the images allow us both to identify people and know how to feel

Bibliography:

• https://cmusatyalab.github.io/openface/
• https://github.com/justadudewhohacks/face-api.js?files=1
• https://github.com/tensorflow/tfjs-core
• https://azure.microsoft.com/en-us/services/cognitive-services/face/
• https://www.ibm.com/watson/services/visual-recognition/
• https://lambdalabs.com/face-recognition-api
• https://cloud.google.com/vision/
• https://blog.rapidapi.com/top-facial-recognition-apis/
• https://www.elmundo.es/papel/historias/2017/12/15/5a327dece5fdea34758b45ef.html