How to Choose the Right Video AI Analytics Supplier
PhD, Global Channel Partner Manager | Head of Sales Europe
The most effective solution to a business problem is where the innovation coincides with the right business case. As an AI algorithm provider, you can create the most complex algorithm, but if it doesn't translate properly into a business case - you will fail. On the other hand, as a consumer of an AI service, if you don't see the time to enrich your offering and product to already properly exploit the potential of AI, you will fall behind your competitors.
In this article, I will talk about the rationale behind how to consciously choose an AI technology provider that, by being aware of its expertise, will help businesses solve their problems in the most effective way. Plus, without losing and even saving money.
As the multi-billion-dollar market for artificial-intelligence-based video analytics continues to grow, so does the number of video analytics solution providers.
This list is in fact far from exhaustive, given that AI-based analytics is gaining interest and becoming mainstream in 2023 video surveillance. Users are expecting more accurate alerts based on object detection instead of motion detection, hardware vendors are developing more powerful but compact chip sets for deployment, and more start-up solution providers are carving out their niche in the market.
If we are a systems integrator, the question arises as to how, in the maze of possibilities, marketing assurances of 100% effectiveness (really?! Never trust that!), to choose the best supplier for us and the end user. Admittedly, the criteria may differ slightly for each vertical market, but there are a few aspects and myriads that I would like to highlight here:
● Open solution ● Integrated ecosystem ● Ease of use ● Durability and operational reliability ● Versatility ● More than good support and integration ● Low total cost of use or ownership ● Competitive advantage ● Cybersecurity
The open platform allows the user to have complete flexibility, avoid being locked into any particular manufacturer, and utilize the best-of-breed solution available in each category. Use existing CCTV architecture, don't replace cameras every 2 years, and give it new life by applying additional analytics. Being camera-agnostic is one of the main factors in choosing the analytics provider within a reasonable budget. Future expansion gives freedom in choosing the hardware and service becomes easier.
It does not matter whether you implement analytics in the cloud, whether it is an enterprise way on your own company's servers or whether you prefer analytics at the edge of the network, the solution should always provide the same efficiency, and the same capabilities. The user experience, and its graphical interface, should be integrated, multiplatform, and browser-based access should be the most universal and accessible.
Ease of use
One of the biggest challenges in the security market is constantly reviewing live cameras from multiple locations. The operator becomes 'blind' after just one hour of effectively, and with tension, reviewing the live view. That is, AI should make it easier for the user to operate security systems. Note whether the handling of any alarms is easy, how effective the model is (FP, FN) and how easily the analytics assimilates into the existing infrastructure. Is there an opportunity to scale it - to use it in larger volumes? How efficient are the processing systems?
Let's look at:
●Easy to set up: just switch ON. That’s it. Plug-and-play saves money. “It just works” should not just be a slogan.●Ease of use: the interface should be intuitive, requiring no special training. Operators should execute the business process, not spend hours learning new interfaces.●Easy to scale: the solution must be designed to seamlessly scale in different ways: number of cameras (e.g., from a few to a few thousand); deployment locations (multi-site friendly), multi-analytics on the same channels, etc.
Durability and operational reliability
Traditional video motion detection-based analytics have many limitations and false alarms, which are now faced by many monitoring stations with the growth of camera connections. AI-based analytics were developed, primarily to identify different objects in the videos with high accuracy in real time.
However, such accuracy must be achievable in different real-life environments. Even in the best installations, we will have the problem of insufficient illumination of the scene, snowfall or rain, the sight of a spider on the built-in infrared illuminators. All of this is something that good AI analytics must embrace.
In the case of behavior recognition, the system must detect the target behavior like slipping, falling, or fights and vandalism acts.
A more robust solution means less time and resources spent on false alarms. A more accurate system can improve safety and security on the premises. We need to understand that the intruder must be stopped at the initial stage when he wants to cross our fence, so a proper mask drawing is always needed.
Are the POC or tests available? Did you check the case studies or references? Make sure you rely on a reliable partner with a mature solution.
Comprehensive, multifunctional video analytics is the most effective choice for system integrators in the long term. It is not just object detection and classification that is important here. AI can be trained to recognize higher levels of detail (e.g. faces, age, gender, number plates), track objects (including people and vehicles) and detect specific human behaviors (e.g. theft, littering, aggressive actions like vandalism or fights). Don't entrust your business to 'one-trick pony' companies. They are in the business of experimentation, not security. Minimize the risks. Are they really specialists in the machine and deep learning? AI is a big area of knowledge and science, so trust the experts. And not solutions that are created as if by the way.
In other words, a more versatile analytics solution can recognize more types and behaviors of objects for more use cases. Most users have certain pain points today and are looking for only one or a few solutions. However, as the organization grows, new situations and requirements may arise, that call for new detection functions in video analytics. Think about your future. Make your business from AI-ready to AI-utilizing.
More than good support and smart integration
One of the main problems with using the latest technology is the lack of specialists. There is nothing wrong with asking for help. Check if your supplier offers you the support you expect, and if they are ready to take on difficult challenges even after working hours (and preferably 24/7!). Is support, and training included in the price of what you are buying? Does the offer include implementation? What is after-sales care like? Check references, and preferably find out for yourself on a small installation how important you as a customer are to the supplier. When it comes to responsiveness, good technical support is part of the ease of use, where the system integrator and the user can be sure that every question can be answered via e-mail or telephone by the manufacturer.
Does the company offer integrations with existing systems? Will the expenditure made on the infrastructure two to three years ago be an expenditure used up or will it only register on the cost side? Such questions are worth answering in advance. Lack of integration with other systems is usually a barrier to entry into the world of AI.
When it comes to having a one-stop-shop solution, good support means that the manufacturer can provide easy integration with third-party systems, which includes API support. One example is access control. Video analytics is a great tool to enhance access security (e.g., facial recognition to open doors for employees; LPR for parking management; weapon detection combined with automatic notification systems).
Low total cost of ownership
Cost is often the deciding factor, especially in the SMB market. Customer expectations are high, and prices for higher resolution cameras and video surveillance equipment are falling, their use is increasing, which means more data to process than ever. A solution that is good today is rapidly becoming obsolete. A seemingly simple replacement of cameras at a facility from HD to 4K resolution can result in the cost of replacing servers (VMS, CMS), network switching equipment, and bandwidth of the overall network. Therefore, an AI solution needs to be designed to make the most of the existing infrastructure without lagging behind new developments.
Good analytics software is not only capable of multiple functions, but its algorithms are also powerful enough to process more data in the same server specification, and it does not require the installation of expensive cameras to have good accuracy, increasing cost savings for the overall system. Always look on the total cost of the installation and ROI.
One of the main competitive advantages of using artificial intelligence in business is its ability to process and analyze vast amounts of data quickly and accurately. This allows companies to identify patterns and insights that may not be visible through traditional methods, and make better-informed decisions based on this information.
AI can also help businesses automate tasks and streamline processes, reducing human error and increasing efficiency. For monitoring stations, AI can provide a competitive advantage by automating the monitoring and analysis of data, allowing for faster and more accurate detection of anomalies and potential issues.
For example, in a power plant, AI can be used to monitor the performance of equipment and detect any abnormal changes in temperature, pressure, or other parameters. This allows operators to take corrective action before a failure occurs, reducing downtime and maintenance costs.
In a security monitoring station, AI can be used to analyze video feeds and detect suspicious behavior or objects. This can improve the effectiveness of security personnel and reduce the risk of security breaches.
Additionally, AI can be used to predict future trends and patterns based on historical data, allowing operators to proactively address potential issues before they occur. This can help to reduce risks and improve the overall performance of the monitoring station.
Overall, the use of AI in monitoring stations can provide a competitive advantage by improving the accuracy and speed of data analysis, detecting anomalies and potential issues before they become major problems, and enabling proactive decision-making.
There are several main concerns of cybersecurity in CCTV installations. Some of the most significant concerns are:
●Unauthorized access: One of the biggest concerns is the risk of unauthorized access to the CCTV system. If an attacker gains access to the system, they can potentially view or even control the cameras, which could compromise the privacy and security of the location being monitored.●Data breaches: CCTV systems can collect and store a large amount of sensitive data, including footage and audio recordings. A data breach could expose this data to unauthorized individuals, leading to privacy violations and potentially even legal consequences.●Malware attacks: Malware attacks, such as ransomware, can target CCTV systems and encrypt or destroy the data stored on them. This can disrupt the functioning of the system and potentially compromise its ability to monitor and record the area being monitored.●Insecure communication: Insecure communication: CCTV systems often use wireless communication methods to transmit data to monitoring stations. If this communication is not properly secured, attackers can intercept and view the data being transmitted.●Lack of updates and patches: CCTV systems require regular updates and patches to address security vulnerabilities and improve system performance. Failure to apply these updates can leave the system vulnerable to cyber-attacks.
Overall, ensuring the cybersecurity of CCTV installations requires implementing robust security measures, such as access controls, encryption, and regular updates and patches. It is also important to train users on cybersecurity best practices and monitor the system for any signs of unusual activity.
Finally, you are ready to choose your AI provider. Go through this checklist and choose with confidence.
At Scylla, we have extensive experience working with various organizations across industries and know firsthand their security and safety challenges. We train our models on large, quality datasets to increase the performance, accuracy, and scalability of our solutions. Scylla performs robust analytics, detecting a wide variety of threats, identifying patterns, and providing valuable insights. It seamlessly integrates with the existing security cameras, infrastructure, and VMS such as Milestone, Genetec, Nx Witness, among others, and is easy to use and scale.
Stay up to date with all of new stories