Source
The Internet of Things (IoT) is the technology that allows connecting everyday objects with the Internet and this is done through small sensors and microchips that, installed on the objects, allow us to interact with them.
With the use of this technology, any of the objects that surround us can connect to the cloud and send the information they obtain from the environment for further processing.
But the continuous growth of this technology and the information generated raises problems such as increased latency in data transmission, as well as security problems for users and data in transactions.
Source
o avoid these problems, what is called edge computing is used, which allows the partial or total processing of data on the IoT device, which minimizes the amount of information that is sent to the cloud.
This allows reducing latency times as small packets of information to be transmitted and reinforces security by having to move less data through the Internet, in addition to relieving the work of the cloud.
In this environment Sony announced a few days ago the manufacture of two models of intelligent vision sensors equipped with AI, which are capable of processing the images before sending them to the cloud.
Source
From a technical point of view, the new sensors are made up of two layers, one with a chip that captures the pixels and another chip equipped with artificial intelligence that processes the captured pixels.
This second chip generates metadata (semantic information related to image data) that it sends instead of the image, which reduces the volume of data to transmit and security problems, as well as cloud saturation.
The chip that captures the pixels is backlit and has approximately 12.3 effective megapixels, which can capture information at a wide viewing angle.
More information:
https://www.prnewswire.com/news-releases/sony-to-release-worlds-first-intelligent-vision-sensors-with-ai-processing-functionality-301059083.html
https://www.eenewseurope.com/news/intelligent-vision-sensors-have-ai-processing-functionality/page/0/1
Versión en español