We are pleased to announce we have partnered with @deepomatic to deploy its Visual Automation Technology across our national network by the end of 2022:

We decided to join 50inTech and to be represented on their Job Fair during their newest Bootcamp : Wannabe a CTO. Come meet our CTO, Vincent Delaitre, during the job fair and discover our job openings in our dev team. https://t.co/3U7xNrBARI

Telecom companies are facing quality issues with the deployment of #fibre infrastructure. We teamed up with Pulse to hear how telecom executives are currently dealing with this challenge. D iscover their views. https://t.co/EcdBWrAxHu

[SAVE THE DATE] MEETUP COMPUTER VISION Hello everyone! The first Computer Vision MeetUp of 2021 will take place **next week**, on Wednesday, January 13th at 7pm online. Registration here: https://t.co/xaT2SIPan7

1st prize in the EIT digital challenge, winner of the i-Nov innovation competition, top 10 computer vision platform in the world, commercial success and new clients… Newspaper Le Figaro looks back on the latest accomplishments of Deepomatic ! https://t.co/JGV2NOMvaC

We are very proud to have been named in the top 20 French AI startups to watch 🇫🇷 🚀 Thank you @Siftedeu and @FranceDigitale ! #computervision #deeplearning

Deepomatic won the 1st prize of the #EITDigitalChallenge 2020 🏆 ! We are very honored that the jury selected us among 400 European deep tech scale ups and recognized our visual automation technology as a major innovation for the market. #EITCommunity #DeepTech #computervision

We have been selected as one of @hellotmrc’s Deep Tech Pioneers! Come and meet us at the virtual Hello Tomorrow Global Summit on 16th-20th November, 2020! ➡️ https://t.co/AjyF0RYghB #DeepTechPioneer #HTSummit

- The 6 steps to create your own image recognition system : https://t.co/lhJ7lixDjD - White Paper on Visual Automation : https://t.co/94KYGFE9yI

That's it, this thread is over. You now know the basics of image recognition. Thank you very much for reading it and if you want to know more, some documents are available here 👇 and in DM.

It's important to know all the possibilities offered by image recognition. This enables companies to determine which technique is best suited to their needs.

I see you coming. Why not always use the most precise technique? Very good question, it's nice to see you follow! Quite simply, the higher the accuracy, the higher the cost and the longer it takes to set up your recognition system.

This technique is notably used in the design of autonomous cars, allowing them to clearly identify road markings, road signs, other vehicles, etc…

This meticulous analysis of each component of an image is mainly used for projects that are both complex and delicate, which cannot be satisfied with approximate precision.

Finally, the main event, the highlight of the show, the most precise technique of all: Segmentation. It enables pixel-precise detection and very precise mapping of each object within an image.

Widely used in video surveillance, detection can, for example, allow you to detect when someone enters a room, or to count the number of people in a room at a given moment.

Detection is much more precise because it enables, in addition to an analysis of the presence or name of the searched concepts, to locate them in the image. And that's not all, it also makes it possible to count the number of concepts present. Convenient, isn't it?

The third technique is somewhat different. Called Detection, it consists in making a bounding box appear on the image in order to locate the element(s) you are looking for.

This technique is very useful for companies needing to categorise their offer. For example, a travel agency can easily categorise destinations or hotels based on the images they have.

The second technique is called tagging, and it is nothing more than an evolution of classification. The process is the same, only instead of assigning a class to the image, tagging puts several labels on it, as you can see below.