Thievery often happens in places like storage areas or depots, as they provide easy access to hundreds of items in one location. Plus, many of these facilities are not adequately secured, allowing thieves to slip in and out unnoticeably. Merely placing security surveillance is often not a good enough solution. Envision having an intelligent system that can trigger a warning every time a suspicious activity is detected. How great would it be?! In this blog, we highlight another successful collaboration with the students. Find out how we built a Proof of Concept for a customer that can be used not only for security purposes, but also for other different use cases using computer vision!

The use case

Leaving valuable belongings in unsupervised depots or storage areas can make some people feel anxious. It is also impossible to be suspicious of every person who comes near the stored goods. At the same time, you can’t help but feel uneasy about it. One of our customers asked us to develop a smart system to help keep an eye on suspicious activities at those unsupervised areas. What an excellent opportunity for our interns to help us create a Proof of Concept and bring the solution into life.

The approach

Unlike the previous projects we did with the students, this time, it involved an actual customer. Luckily for them, they still got the freedom over their creativity. As we still held the same belief of having them totally “own” the project. Only this time, they got to work even more closely with experts from Mediaan Conclusion, to make sure that everything went smoothly. At the same time, we also actively helped them develop & improve their soft skills, and to work with the Agile methodology.

The project

The goal of the project was to create a Proof of Concept using the principles we would like to use for the AIVI platform. AIVI will be a Mediaan Conclusion product, that combines the power of computer vision, our own detection algorithms (which we will integrate to the platform in the future), and the flexibility to be linked to already existing systems without too much of a hassle. During this project, pre-recorded camera feeds from different angles of the storage areas were available for use. Though they were pre-recorded, we made sure to build a PoC capable of detecting suspicious activities in real-time.

The result

A team consisting of Mediaan Conclusion experts and students built an application that has multiple functionalities:

1. Detecting objects near the critical zone

The first functionality is to capture any “object”, whether human or anything that enters the focus area. It would then analyze how close these objects would come in the zone, where the valuable goods are located. “Not close? No problem, have a nice day.” “Pretty close? I’m keeping my artificial eye on you!”

2. Calculating the speed of the moving objects

Merely detecting an object is not enough to trigger a warning, as it would constantly go off. Thanks to the clever use of short-termed historical data (a few seconds back, up to a minute), it is possible to also calculate the speed at which the object is moving. Faster moving objects could mean that they are only passing through the area, even if they are a bit close to the critical zone. On the other hand, slower moving objects, or those that are standing still, might trigger a warning.

3. Fully customizable

We used the term “might trigger” before, because it was fully customizable to lower or raise the danger threshold to trigger alarms. The reason behind this is to fine tune the program, so it can be used on different sites where other variables might come into play. Whatever the trigger might be, a suspicious person, or an obstructed view, a warning would be triggered to send different kinds of messages, like email or text, to the concerned parties.

4. Filter function

The recorded images are stored in a database that has filter functionality. This way, you can filter warnings and/or no warnings and strange behavior to be reviewed, which makes it easier and faster for the users to take the next action.

Conclusion

Once again, a successful project was done together with the students. Not only did we contribute in cultivating future tech talents, but we also got fresh new ideas in return. The PoC was built in a way that it is easy to be reused in different circumstances, for different use cases. We will for sure expand this PoC, not only within Mediaan Conclusion, but work it further with the customer who asked us to develop this product. Do you also want to step-up your security camera’s ability using computer vision? Our experts are always ready to help you! In case you missed other interesting projects we did with the students, you can check them out here:

This blog is written by Sébastien Bodart – Consulting Project Manager and Internship Supervisor at Mediaan Conclusion