AI made in Africa for Africa

AI made in Africa for Africa

Headquartered in Mauritius and owned by Convergence Partners, inq started life as a telecoms company in Africa and has since moved into the cloud and digital services arena. It currently has a presence in Johannesburg, Gaborone, Lusaka, Blantyre, Lilongwe, Lagos, Abuja, Port Harcourt and Abidjan, among others.

  

While the company has a number of operations on the continent, its Johannesburg office is more of a point of presence through which it operates in South Africa through its partners. Hi-Tech Security Solutions spoke to Pramod Venkatesh, the Group CTO about what the company offers in the security space.

With its goal of supporting customers in their digital transformation, inq offers a range of services, from intelligent connectivity, local and public cloud connectivity services and more pertinent to Hi-Tech Security Solutions’ readers, AI analytics services for surveillance.

 


Pramod Venkatesh.

 

Venkatesh explains that part of the company’s transformation into the digital services arena was a focus on developing video analytics based on AI. The company develops its own analytical engines, in Africa, specifically designed for the realities of the continent. As such, it realised from the start that video surveillance in Africa is different and most companies are not in a position to install the latest and greatest (and most expensive) cameras and management platforms and it would therefore have to cater to that reality in its solutions.

Any camera will do

inq’s analytics are overlayed over almost any camera’s stream, even some of the cheaper ones, providing customers with reliable AI services. Venkatesh reiterates that this is because the solutions were developed from the ground up for use in Africa.

The analytical functions on offer at the moment include anything from line crossing to mask detection, LPR and facial recognition, object detection and more. The list of functions is continually advancing and inq is also able to cater for specific development requirements as it owns the software. 

As-a-service or as you like

The solutions are designed and sold as a cloud-based service, targeting a growing list of verticals, including telecoms, retail and industrial settings. And while the service runs on inq’s own cloud management platform as a ‘cloud-native’ application, meaning it runs in a browser without requiring the installation of additional software (basically an AI-enabled NVR in the cloud), Venkatesh notes that it is also API-enabled to allow customers to integrate the services into their existing management platforms.

For customers that want onsite infrastructure, inq also makes this available, supplying the servers or providing the customer with the server specifications required (it must have a graphics card, for example). The customer then pays one perpetual licencing fee along with an annual maintenance fee for upgrades, improvements and new analytics. 

The cloud-based services model runs on an operational expenditure (opex) model with a monthly fee per camera. The Basic package includes most of the analytics services, while the upgraded model includes recognition analytics – LPR and facial recognition. This is available at about a 25% premium on the Basic package.

The facial recognition AI was trained on a dataset from Africa to ensure its accuracy on the continent and the AI is always learning and improving. 

The management platform itself has been designed to be easy to use and configure, making setting rules or running specific analytics on a camera simple; these are built in and include line crossing, people counting, heat maps, facial recognition (companies can load pictures of their employees and the system will automatically incorporate those images into its face processing), tampering detection, mask detection, region of interest and more. The analytics can be run on live streams or stored video and video can be uploaded as a file to be analysed if required.

When it comes to object detection, the system is currently able to recognise 80 objects, but this number will increase. An example of a past addition would be adding snakes to the list of objects that can be detected, possibly a unique African requirement.

Notifications built into the system include raising alerts on the management console (in the browser) or via SMS, WhatsApp or email. Also included is reporting functionality to provide management with information via a range of templates, while other reports can be built as required.

inq will be making more announcements about its future developments in South Africa in the new year. 

For more information contact Prince Cekeshe, inq, princec@inq.incwww.inq.inc

 

Article source: http://www.securitysa.com/14689r

Image source: http://www.securitysa.com/14689r