How AI is Igniting Video in Enterprise and Industry
Go back twenty years and the term “industrial video” meant terrible corporate videos produced on amateur video kit by enthusiastic internal corporate communications departments. Today, though, 5G networks, 4K cameras, and AI platforms are transforming the role of video in enterprise and industry in ways that nobody could have predicted to make businesses more efficient, safer, and improve their customer experience.
Let’s look at the three components that enable intelligent video applications:
The Three Components of Intelligent Video
First you have the intelligent camera. This camera will capture images in Ultra High Definition 4K video. It will also produce metadata frame by frame that describes the objects that it sees. These objects can be faces, clothes, cars, number plates, bags, anything that the camera’s software has been programmed to detect. This data needs to be generated in real time so it can be stored and acted upon in real time which is why it takes place at the point the image is captured in the intelligent camera.
Second, we have the AI platform that comprises a database that stores all the metadata produced by the intelligent camera, a learning module that makes sense of the metadata, and an inference engine that creates outcomes based on that learning. A typical outcome would be a successful match by a facial recognition system. The AI platform should be provided as a cloud service that integrates through standardised interfaces with the enterprise’s other IT systems that both provide it with data and can be remotely controlled.
And third, we have the network that connects the intelligent camera to the AI platform and all the other integrated IT systems. This network must carry huge amounts of video and metadata in an instant so historically in enterprise it would be a wired or fibre network around the campus or site. Wired networks limit the camera locations and are expensive to maintain so increasingly this network will be a 5G mobile network that offers both the capacity and speed without limitations on location and it is the availability of 5G networks that will enable intelligent video applications in many industries.
So, let’s look at some of the emerging applications of intelligent video in different industries. In general, AI is used to achieve one of two outcomes. It either makes the business operations more efficient and hence the business becomes more cost-effective to run. Or it improves the business’s customer experience. In the old days, project managers used to talk about the three interlocked attributes of a project that could be improved: cost, quality, or time. All three attributes could not improve simultaneously. The model then was that you could reduce the cost of a project and maintain quality but the time to deliver would go up. Or you could reduce the time to deliver a project and maintain the cost but the quality would go down. Bringing AI and its associated automation to enterprise applications raises the possibility that we can improve all three attributes simultaneously and get things done cheaper, better, and quicker.
The Emerging Applications for Intelligent Video
Health and safety on construction sites is being greatly improved by intelligent video systems that can detect and alert when somebody enters the site without wearing a hard hat. Connect the system to the entry gates and you guarantee nobody can enter this hazardous environment without the appropriate safety gear.
In this context, the campus is not just the name given to University sites. Any large manufacturing unit or sports stadium or collection of offices such as Huawei HQ in Shenzhen can be considered a campus. In the old days, security guards would patrol the perimeters. If security was particularly tight then there may be physical trip wires which were later replaced by laser beams but still called trip wires. Today, Intelligent Video Surveillance systems let you define invisible trip wires or hot zones on a screen that produce an alert when breached with nothing physical on the ground at all.
High-resolution video conferencing systems are increasingly being used to create virtual classrooms and bring the expertise of top teachers and lecturers to students in remote locations. These systems are a far cry from the old television broadcasts of dry university lectures in the 1970s. They are fully interactive teaching experiences with lecturers and students able to see and interact with each other as though they were in the same room. Adding AI to this is a system that converts the attributes of the students’ faces into data that is interpreted into an engagement metric and lets the school know if the individual students are interested, happy, bored, or paying attention.
For many years the traffic cops used speed cameras that could detect how fast we were driving and take a photo of anyone exceeding the limit. Today’s intelligent equivalent can take pictures through the windscreen and detect if you are wearing a seat belt, or talking on your mobile phone, or otherwise breaking the rules of the road. The fine will be waiting in your email inbox before you finish your journey.
As well as automating the robots on the production line, AI is increasingly being used in the final step of the manufacturing process to control product quality. AI takes video images of the finished product and determines if there are any faults or abnormalities in its production. It can do this more efficiently and consistently than the human eye. Using intelligent video in this way is being used in everything from precision manufacturing to domestic pizza delivery.
Recycling of household waste is a priority in many countries but laws and coloured bins are only as effective as the humans they are targeted at. Waste management companies are now able to use intelligent image recognition and robots to sort trash at the depot into different categories of waste (such as food, aluminium, plastic, or paper) for effective recycling.
There’s probably no other public location with higher security requirements than airports and in some countries railway stations have equivalent requirements. We are already running trials of boarding planes by facial recognition without the crew checking your boarding pass and passport. It happened to me at LAX last year. In the future, facial recognition will be used end to end in the airport from the roadside drop-off, through baggage drop, security, and immigration to speed passengers quickly and securely through the airport.
A far cry from the old CCTV systems, we can see that the use of intelligent video across enterprise and industry is set to boom and it’s enabled by 5G networks, intelligent 4K cameras, and AI platforms in the cloud.
Here at Huawei, we make all three.