Get best of the week

Visitor Detection and Counting: Cloud vs. camera counter

image

In retail, to measure the effectiveness of the process of transforming an ordinary visitor into a paying customer, a conversion indicator is used based on the data of counting visitors.

Previously, systems operating on the infrared ray or on thermal sensors were deployed to analyze and collect statistics. Today it’s enough to install at least one camera connected to the video analytics service.

It would seem that a face recognition system that transforms any camera into a “counter” of unique faces should once and for all close the topic of tracking people. But specialized equipment for detection and counting has not disappeared from the market and is represented, inter alia, by devices from Ivideon and Dahua.

This material will help resolve the paradox.


image

The Ivideon Faces face recognition system analyzes the video stream from the camera at the entrance and helps to find out the number of unique and new visitors at the facility, disaggregated by day and hour. Cloud face recognition works on cameras with Ivideon firmware, or on cameras connected via Ivideon Server.

The service provides information about the time of day when people visit the outlet, demographic characteristics (gender and age), and even about emotions on their faces. The solution is easy to use and remote maintenance.

Another method of cloud computing is not directly related to face recognition and is implemented on the basis of other video analytics algorithms - you can use any camera to count visitors by connecting a tariff with the function of a cloud visitor counter to it.

image

However, some of our customers, for example, the Moscow Jewelry Factory chain of stores, chose a specialized device - Ivideon Counter 3D . The face recognition system, as the name implies, works specifically with faces and is not always effective for counting a large number of visitors.

A person can easily obscure the face of another visitor, and then the system will count them as one person. To increase accuracy, the camera lens should be at a height slightly higher than human height and be aimed exactly at the incoming stream, which is difficult to implement in a real store. Moreover, such a solution is effectively considered only by a one-way flow of visitors.

Counter Camera Ivideon Counter 3D



image

The counter, built on the basis of a 1.3-megapixel camera with two sensors, is installed vertically above the aisle so that all incoming and outgoing visitors get into the lenses. Covering the control zone at angles of 91 ° (horizontal) and 66 ° (vertical), Ivideon Counter 3D counts people with an accuracy of at least 95%.

The location of the lenses next to each other allows each sensor to capture images of the same area at slightly different angles, which increases the perception of depth and helps to filter objects by height. Technology gives the camera the ability to distinguish between people and other objects (such as carts).

The camera monitors visitors crossing a given line and monitors all pedestrian traffic. Line transitions to one side and the other are considered separately - users, accordingly, are available several types of reports sorted by type of traffic: input, output, or all together.

image

Data in reports can be sorted by date and with accuracy in time - from several hours to months. Moreover, for each counter separately, which facilitates the collection of statistical data in a room where there are several inputs and outputs and several "counters".

Ivideon Counter 3D is based on technology from Ivideon's strategic partner, Dahua Technology . In 2019, Dahua mastered the release of new counting cameraswith two lenses. As part of the partnership integration program, Dahua uses Ivideon cloud service on meters.

Ivideon Counter 3D runs on the native Dahua protocol and has almost all the capabilities of other meters released under the Dahua brand. However, working with the native protocol provides unique opportunities - for example, an API for the counter. Now a third-party system can query to receive statistics on visitor access from the Ivideon cloud.

Face Recognition vs. smart meter


image
Ivideon Faces system (“Faces” section in your account)

The face recognition system works effectively within the access control system, in the cash zone, to identify VIP or potentially dangerous visitors to the store. The system determines the movement of visitors throughout the room - you can easily determine the areas that customers visit most often, where the lines are formed and where they spend the most time.

The counter camera measures pedestrian traffic at the entrances to the store, in the area in front of the storefront or at specific points near the shelves. The strategic advantage of the meter compared to any other cloud camera is the ability to work without a permanent connection to the Internet and the cloud.

Counter 3D works effectively in the event of a network break - despite the lack of communication, the device will count visitors and write data to a local archive. After restoring access to the Internet channel, all statistics are “downloaded” to the cloud and become accessible in the user's personal web account.

image
The counter data is available in the “Visitors” section of the Ivideon personal account. The

vertical arrangement of Ivideon Counter 3D provides a stable and clear background for image analysis. If the background is a complex scene, for example, when shooting on a horizontally located camera, the detection efficiency of moving objects is reduced.

The camera must be installed at a distance of at least 2.5 meters above the target. The recommended width of the exit and entrance should be within 3 meters - to calculate statistics at the entrance, the width of which exceeds 3 meters, it is recommended to install several cameras.

Ivideon Counter 3D can work in low light conditions, low contrast, low or high brightness of external light, thanks to the support of image enhancement technologies:

  • Back Light Compensation - the function compensates for excess lighting;
  • 3D Digital Noise Reduction - digital noise reduction;
  • Digital Wide Dynamic Range - the use of the extended dynamic range and high digital signal processing allows you to get a high-quality image of both bright and dark parts of the frame.

By collecting data on pedestrian traffic during the busiest hours of the day, managers can withdraw a sufficient number of personnel during peak hours. The data allows you to optimize the layout of the room to provide an easy and comfortable service. But most importantly, you’ll get a tool for calculating conversion rates.

Learn and use conversions with a visitor counter


The practice of calculating the cost of one attracted buyer in offline retail is not as common as online. However, an analysis of how effectively the store works out every opportunity to attract a potential customer will help the business increase its profit. To do this, it is enough to add the cost of the rent, the costs of window dressing and advertising, and then divide the resulting amount by the number of visitors.

The level of conversion of visitors to customers in general is considered in two ways:

  1. (Number of buyers / Number of visitors) * 100%;
  2. (Number of customers / Number of visitors * Error rate) * 100%.

The margin of error correction should take into account customers who do not have a purchase need (they went in to warm up or accidentally ended up in a store).
Settlements according to the first formula with 100 visitors per day and 50 purchases will give a conversion of 50%. Of course, it would be nice to separate potential buyers from casual visitors to the outlet. In this case, the conversion will be even higher.

But what is the exact formula to calculate and how to determine the error? In fact, this is not so important. Even for different types of businesses, the conversion rate is significantly different. For a car salon, a 3% conversion will be a very good indicator, and for a stationery store, a complete failure.

Given the imperfection of all calculation methods, monitoring the relative numbers remains the best option. What matters is not the absolute numbers themselves, but how they relate and how they change over time. You can choose any option - the relative increase or decrease in conversion will look the same.

Thus, if, according to one method, your weekly conversion has increased from 20% to 25%, and according to another, from 40% to 50%, then in both cases the increase will be 1.25 times. It is important to consider statistics for a sufficient period of time for analysis. Convenient statistics from Ivideon Counter 3D , available both in your online dashboard and for analysis in other systems by API, will help you make the most effective decision for developing your business.

More articles:


All Articles