People Counting Technologies: A Comprehensive Guide
The need for reliable and accurate people counting technologies for businesses is now more critical than ever. In retail, for example, various technological solutions are implemented to ensure that stores’ operations run as smoothly and efficiently as possible. The technology that goes into business operations is vastly versatile and covers every aspect, from security cameras and POS systems to RF and RFID security tags.
With consumer behavior shifting due to the pandemic, retail stores are looking for ways to optimize their conversion rates. To begin the optimization process, businesses need to gain insights into their visitor data to understand shopping behavior – this is where visitor counters come into play.
Customer counters provide actionable foot traffic data to help managers make informed business decisions. They can be used in different industries like retail, shopping malls, supermarkets, hotels, casinos, airports, libraries, schools or universities, gas stations, banks, museums, theme parks, or any other business.
While people counters are relatively simple, it can get overwhelming as there are many options out there to choose from.
Beam sensors consist of receiver and transmitter units that are installed side by side on the entrances. When the transmission signal is blocked due to an object obstruction, a count occurs. Beam sensors have some downsides; For example, they don’t provide in & out numbers separately since they have no sense of direction. Also, beam sensors aren’t the most accurate since side by side objects are counted as one. The accuracy decreases as the door width increases. Lastly, beam counters don’t filter items like shopping carts or baby carriages.
Thermal counters use a person’s body heat to measure footfall traffic. They create images using infrared radiation similar to a typical camera that forms an image using visible light. Thermal sensors are installed top-down to detect human temperature on entrances to count them.
Typically, thermal sensors are negatively affected by sunlight since sunlight covers the whole spectrum of light.
Crowded groups can also decrease the accuracy since thermal counters cannot differentiate between objects due to their temperature signatures.
When the environment temperature is close to human body temperature, they can differentiate objects from humans, but not children from adults. Thermal sensors can count in no light conditions, but more sensors are required due to their low field, thus increasing cost.
2D Mono Counters
2D Mono (Monocular) sensors use a single camera lens for counting. Mono counters are installed top-down to detect moving objects only. The algorithm digitally removes the static background and only tracks the moving objects. Since mono counters lack vision depth, there is no smart object detection algorithm, and every moving object is counted. Due to this, they work only in low traffic areas where the lighting is bright and consistent. They are susceptible to count wrong in environments that contain shadow, sunlight, crowded group entrances, baby carts, shopping carts, and children. It has a high field of view.
Mono counters are cost-effective and easy to install, but they provide inaccurate data that is dependent on the environment.
Wi-Fi counters vary in size but can be as large as a home router. They work as long as the Wi-Fi is enabled and there is access to it. This can result in inaccurate results since this type of sensor is dependent on having a Wi-Fi access point. Wi-Fi counters can be complicated to set-up and are best used as a secondary counter and not as a replacement for a primary system.
Time-of-Flight counters are similar in size to thermal cameras. This counting method is based on measuring the time difference between a signal’s emission and its return to the sensor. Time-of-Flight sensors work by sending a signal to objects beneath the sensor and then recording the reflection that bounces back to register a count. Several types of signals measure the distance between a sensor and an object; these are Ultrasound, Infrared, and Laser signals.
Time-of-Flight is old technology. Since sunlight covers the whole spectrum, it can significantly disrupt the processed signal. It is known to provide wrong information due to the signal quality that decreases as the distance between the object and the sensor increases. In crowded entrances, performance can be poor as ToF sensors cannot differentiate objects from visitors or adults from children. Time-of-Flight sensors can work in total darkness but require multiple sensors to cover a wider area, increasing the overall cost.
3D Stereo Counters
3D Stereo vision is the extraction of 3-dimensional information from images. Related images can be obtained by a biological unit such as the human eye or an artificial unit such as a camera.
In humans, stereo vision works when each eye captures its view, and two separate images are sent to the brain for processing. When the two images arrive simultaneously in the back of the brain, they are merged into one picture. The brain combines the two images by matching up the similarities and adding both pictures’ small differences. The slight differences between the two images add up to a big difference in the final picture. Thus, three spatial dimensions are created; width, height, depth, or x, y, and z.
An analogy can be made between human stereo vision and computer stereo vision. In computer stereo vision, receptor units are cameras as opposed to human eyes. The same process applies: Two cameras capture two separate images. They are processed simultaneously and combined into one image to provide spatial depth information.
Stereo people counters are installed top-down at any entry/exit points. Stereo counters have vision depth, which allows the cameras to exclude objects that don’t meet the height requirements specified during the calibration process. The overall accuracy of stereo counters can vary from one system to another.
3D Active Stereo Vision
Our 3D active stereo vision technology processes the combined images and creates depth maps to provide accurate and reliable counting. Sensors are installed on the ceiling to monitor the entrance of the location.
In this example, we can see a sample location entrance with 3-dimensional depth information. The heads of the visitors are marked with a red circle. The algorithm tracks this circle. Counting occurs whenever it coincides with imaginary lines that are configured by V-Count support personnel. Since objects are differentiated by their heights, items located side by side can be counted correctly. This technology has a high field of view to cover the area, thus decreasing the customer’s total cost.
V-Count’s Ultima AI: The Ultimate People Counting Sensor
Ultima AI is the next-generation people counting/tracking sensor. With its advanced capabilities, it offers an unmatched accuracy up to 99.9% and uses the most advanced 3D active stereo vision technology and AI-based target tracking with an HD resolution of up to 1280 x 720 active stereo depth.
In addition to more state-of-art features like auto-calibration, night vision, an outstanding wide field of view, and many more. Ultima AI also combines all the top solutions from people and zone counting to queue measurement and demographic analysis into the thinnest people counting/tracking sensor in the world.