Irida Labs is collaborating with Hikvision to overcome the challenges of Vision AI

Irida Labs is collaborating with Hikvision to overcome the challenges of Vision AI

Vision AI - which processes video images within a camera to produce analytics and trigger actions - has enormous potential for sectors such as manufacturing, logistics and smart cities. The significant technical challenges associated with Vision AI have made it hard, if not impossible, for many systems integrators and resellers to deliver. Now, new software, coupled with Hikvision cameras which can process AI algorithms 'at the edge', means that these challenges can be overcome.

In recent years it has become impossible to avoid the excitement surrounding Artificial Intelligence (AI) with tools such as ChatGPT which works with text inputs. For Hikvision Solution Partner, Irida Labs, there is another form of AI which they call 'Vision AI'.

Here, a video stream from video camera or camera module is processed by an onboard AI application which analyzes the visual data and produces an output in the form of analytics data or triggers. In this way, Vision AI uses AI to transform a video camera into an AIoT sensor or monitoring device.

"It's a new way of thinking about how video cameras work and what they can be used for," says Thomas Charisis, Product Growth Director for Irida Labs. "Instead of a human watching a video stream, an onboard AI application can monitor it, interpreting the images according to the AI algorithms, making decisions, and acting accordingly. Without the need to constantly watch the video, the AI application replaces mundane human effort by discarding nearly all unnecessary data before it even leaves the camera."

Overcoming the problem of Vision AI complexity

The vision of Vision AI is extremely compelling. But systems integrators and video hardware resellers soon discovered that such solutions could not be delivered without the help of expert AI programmers. "Most companies who wanted to provide Vision AI solutions to their customers, found it was impossible to deliver," says Thomas. "It was becoming a service that only those with the appropriate in-house programming skills could provide, involving navigating a fragmented and multi-disciplinary ecosystem involving cameras, system design, data campaigns, AI/ML models, edge software optimization, edge device management, and more. For the rest, therefore, instead of being an opportunity for their businesses, Vision AI was becoming "a resource-heavy project, with limited potential for scaling up."

The Irida Labs's Vision AI platform, PerCV.ai, addresses the challenge of simplifying Vision AI in three discrete stages.

1. Off-the-shelf AI apps - or a customized app creation service

The first challenge was the development of the AI applications. The company has created several suites of apps which designed for Industry 4.0 and Smart City sectors. Alternatively, Irida Labs provides a custom AI development service to customers all over the world. Some of the niche Vision AI apps they have developed include inventory monitoring, quality inspection, vehicle monitoring, ANPR, people occupancy for indoor and outdoor spaces, QR code scanning, laser process monitoring, and many more.

2. Deployment of the AI app onto Hikvision cameras

Irida Labs has created a new version of its software platform, PerCV.ai, dedicated to Hikvision cameras. The platform enables non-expert systems integrators and security installers to deploy sophisticated, ready-to-use AI models directly onto Hikvision's HEOP 2.0 devices, such as the DeepinView Series cameras, with just a couple of clicks. Crucially, the system also includes a simple and highly intuitive way for non-experts to develop custom apps to work in unique use cases.

3. Edge processing within the cameras

The PerCV.ai solution is designed to take full advantage of the powerful processing capability that is built into Hikvision's HEOP 2.0 cameras. It means that the video images are analyzed by the camera itself and the only data that needs to be communicated are analytics and triggers that lead to further actions.

With their unique approach, Irida Labs has managed to take all the complexity of AI development and deployment and turn it into an extremely simple workflow. "Our deployment of off-the-shelf solutions is unique in the market", says Thomas. "But it is important to note that what we have achieved is possible thanks to the hardware that we work with. Hikvision cameras have on-board processing capabilities which enables the AI processing to occur within the cameras themselves."

Also critical to the development of the PerCV.ai solution was the Hikvision Embedded Open Platform (HEOP 2.0) interface. This enables technology partners, such as Irida Labs, to develop and run their own applications on Hikvision's hardware. These applications suit the specific needs and unique installation scenarios of customers. According to Thomas, many systems integrators that they work with are already familiar with the HEOP 2.0 interface, and the analytics that can come from it. "In a sense we are simply adding more value within this well-known ecosystem."

The significance for resellers

Many of the client companies that could benefit from Vision AI applications have little, or no technical experience of AI. They might be manufacturers, or a local authority, or even a small business needing a niche application. In all cases they rely on their technology suppliers - systems integrators and security resellers - to provide solutions which address the business challenges that they face and, hopefully, give them a competitive advantage.

For them, Vision AI is a new, and highly powerful, technology that can add significant value to their clients and create new revenue streams for them in the process. There is clearly capital revenue to be gained from selling the AI apps and the hardware that they run on. But in addition, each system needs to be individually customized; the cameras need configuring, and the software needs to be deployed and individually trained in every installation. Moreover, as the customer's needs change, so the Vision AI solution needs expanding, recalibrating and retraining.

Now that the challenges of Vision AI have been overcome, this revenue potential is now an opportunity for all.

Find out more

For more details about the unique Irida Labs and Hikvision partnership, please visit this page. Alternatively, click here for information about the HEOP 2.0 interface which enables Irida Labs to develop and deploy its solution.

Related Links

Hikvision Embedded Open Platform 2.0 (HEOP)

Subscribe to newsletter

Subscribe to our email newsletter to get the latest, trending content from Hikvision