AI: managing cost and maximising ROI

Posted On 26 Jun 2020
Comment: Off

Technology is an ever-changing beast and, while there are always new opportunities to innovate, one also risks getting left behind. Artificial intelligence (AI) is one of today’s most discussed technologies, meaning businesses are rushing to implement it so as to avoid missing a trick. In particular, within security and threat detection, AI-powered solutions are gaining popularity as enablers of unprecedented efficiency and precision. This has compelled organisations to turn to vendors and integrators and demand to trade in their existing video surveillance systems with new ones that leverage AI.

However, businesses should be aware that switching to AI-enabled solutions simply cannot be a straight swap – there are added costs and technical requirements they must take into consideration. Here, Kevin Waterhouse, Managing Director at VCA Technology looks at why the cost of AI, particularly within security, must be considered, why it’s worth overcoming this hurdle and how it can be done.

Viral impact

Until recently, the world seemed to be running at breakneck speed and technology was propelling it forward – then COVID-19 happened. As a consequence of the pandemic, people’s lives have ground to a halt – although, in some areas, they’re returning to normality. Customer demand is wavering unpredictably and this has triggered a financial crisis with repercussions across all industries. Therefore, companies worldwide are re-evaluating upcoming planned investments in order to save funds and stay afloat.

For instance, Gartner reported that 62% of finance leaders are planning cuts to a variety of business functions including risk, facilities, R&D and procurement due to Coronavirus-related disruptions. This doesn’t mean all projects should be postponed or dismissed – after all, innovation is precisely what will enable businesses to agilely adapt to the unfolding situation. However, new investments should be chosen with extra care, paying attention to what new challenges have arisen due to recent events and what technology can cost-efficiently help to tackle them.

The price of accuracy

Video surveillance is one of the areas where artificial intelligence has triggered the most beneficial advancements. The advent of video analytics systems helped monitoring staff battle false alarms, which wasted time and resources – AI has gone a step further and made threat detection even more accurate. In particular, Deep Learning enables systems to, quite literally, “learn” what a potentially dangerous event looks like, ensuring the alerts generated by the tool flag actual security risks. It seems almost counterintuitive to adopt a video analytics surveillance solution that lacks AI capabilities.

However, while the advantages unlocked by AI are undeniable, they come at a price. The processing power required to support this technology is huge, and the implications in terms of increased cost cannot be ignored. Far from being a plug-and-play scenario, AI-enhanced video analytics, for instance, requires a switch from a CPU architecture, whose power is considerably limited, to a custom GPU model, which can support Deep Learning algorithms.

Additionally, businesses should be cautious when it comes to could-based AI services, which are becoming more widely used due to the perceived lower investment. In truth, this model works with a monthly subscription which, on the long run, adds up to a significant cost. Site-based AI solutions are arguably more sustainably-priced and present fewer risks.

This by no means indicates companies should not consider AI-enabled surveillance – it simply reinforces the need for a more practical and price-aware approach. The obvious remedy is seeking a cost-effective AI surveillance tool, ensuring it can quickly drive a significant return on investment.

Cost-effective compliance

Another issue COVID has created for business premises is occupancy: social distancing regulations compel individuals to a safe distance from one another, posing obvious challenges in stores and other commercial locations. While security staff can help, companies need more streamlined and systematic ways to comply with this rule, hence the need to leverage AI and video surveillance.

Depending on their budget, existing equipment and time requirements, companies have a series of options. Pose estimation, which analyses people’s gait, can count people in a space with impeccable precision, ensuring occupancy limits are not surpassed. Like Deep Learning, this technology requires a GPU architecture and demands a specific server on the hardware it uses. Existing PC’s, for instance, will need to be upgraded, which IT teams can support with. Clearly a more complex (and costly) installation, gait analysis is ideal for companies that have more time and a bigger budget but still want bang for their buck, in the form of superb accuracy. Gait analysis can also be an incredibly useful investment for in-store surveillance as it can identify slip, trip and fall incidents, based on customers’ posture, and immediately call upon security staff, before the situation escalates; simply by tracking if someone is prone or upright.

Alternatively, businesses looking to address the occupancy issue with a quicker turnaround can use analytics – without gait analysis capabilities – to count customers in and out of a store and make sure they keep a safe distance. In the best of cases, this can be deployed in the space of two weeks, promptly enabling ROI for the purchaser. Both of these solutions can function with existing cameras, as long as these are placed correctly, so businesses should consider when and how new equipment can be installed, in order for the cost of the process to be contained. It shouldn’t be forgotten that these ‘simpler’ provisions can help with business decisions too – for example, with analytics they can identify store hotspots or queue issues; helping to add value from the off.

In light of recent events, technology investments have never been under closer scrutiny – yet evolution has never been more necessary to adapt to unpredictable and constant changes. At a time when optimisation is fundamental to survive, video surveillance, accompanied by AI and other modern tools, can empower businesses to maximise the cost-efficiency of their investments, so it’s time for them to carefully analyse their needs and means, and choose the path best suited to them.