Frequently asked questions

Due to the cumulative error in a counting solution, the accuracy of the overall system decreases over time, reducing the reliability of data on parking lot occupancy.

Moreover, critical insights, such as the average parking duration or parking violations, cannot be captured.

An example calculation: A parking lot with 500 spaces, which is fully utilized twice a day on average, experiences about 2,000 entries and exits per day (500 spaces × 2 entries + 2 exits). With a typical event accuracy of 98%, this results in 40 miscounts per day. Over the course of a week, the error adds up to 280 miscounts, representing a significant deviation.

Only through constant manual resets — a considerable cost driver over the duration of such a project — can reliable data on parking lot occupancy be maintained.

This issue applies to all counting stations, regardless of detection technology. Nevertheless, SONAH delivers an exceptionally high detection accuracy, ensuring precise results with minimal deviations even for complex counting setups in parking lots or garages

a) Elevated position and unobstructed perspective

b) Continuous power supply

c) 4G or 5G connectivity

a) Each sensor is equipped with a computer that runs our image analysis algorithms (Machine Learning).

b) All calculations and image processing take place exclusively on the sensor itself.

c) Images are processed in real-time and are not stored on the sensor. Processing occurs exclusively on the sensor.

d) During the detection process, no images are transmitted to the cloud, and no analysis of image material is conducted in the cloud.

e) The sensor only sends metadata (e.g., GPS location, timestamp, status, ID) to the backend.

f) There is NO post-processing of images in the cloud, unlike many other "Smart CCTV" services where video streams are sent from the camera to a network device.

The SONAH sensor architecture safeguards user privacy by only storing the data that is strictly necessary.

Personal information is never collected, and unnecessary data is discarded immediately. During the operation of the system, there are four primary types of data transmission, outlined below:

(ROI = Region of Interest = A specific part of the camera image where application-specific recognition algorithms are applied).

a) The number of parking spaces

b) A site plan including the location of poles and streetlights

c) Confirmation of whether batteries need to be provided for daytime power supply to the sensors

d) A perspective video of the site showing all parking spaces, poles, streetlights, etc.

e) A Google Maps link or the exact coordinates of the site

This depends on the local conditions and the installation height. We differentiate between open parking lots, parking garages, and on-street parking spots.

Under optimal conditions, a single sensor can digitize up to 72 parking spots simultaneously.

Additionally, the minimum installation height should be at least 3.5 meters. At a height of e.g. 3.5 meters, approximately 8 on-street parking spots can be digitized. Moreover, at this height, the sensor is protected against vandalism and damage.

Edge computing means that data processing occurs "on the edge," directly on the data device, so only metadata is transmitted. Edge computing is particularly suitable for use with sensors.

Advantages:

• Low roaming (less cost)

• Lower latency (faster response times)

• Preservation of data privacy and capture of only relevant data

• Reduced bandwidth requirements

Disadvantages:

• Reduced computing power compared to cloud computing

• More complex AI development

SONAH VisionSensor with Edge Computing:

1. Data processing occurs directly on the sensor.

2. No personal data is collected or stored.

3. Low data transmission costs (approx. 100 MB per month).

4. Real-time data.

5. Minimal dependency on bandwidth or network availability.

6. Wireless data transmission is possible.

7. Expandable to additional use cases, with costs for processing per use case not scaling.

CCTV Cameras with Cloud Computing:

1. Recording and streaming of images.

2. Requires dedicated lines and extensive infrastructure work.

3. Usable for security applications (security cases).

4. Access to recorded footage is restricted to trained personnel.

5. High computing power in the cloud for AI-based analysis.

6. Reliable data storage through redundancies.

7. Expandable to additional use cases, but costs for processing per use case scale. This means that the more detections are made, the higher the monthly costs in the cloud.

8. Not suitable for public space, only for private property.

The sensors are designed for a minimum of 5 years of operation. Additionally, we offer a 2-year manufacturer warranty.

The lead time depends on the volume and schedule of the project. Typically, lead times of 4 to 6 weeks are achievable without issues.

The installation of the sensor usually takes about 15 minutes. Afterward, the learning system calibrates itself for approximately 4 weeks to adapt to the local conditions (such as the angle of view and lighting conditions), ensuring excellent performance with over 98% accuracy.

Our sensors can be installed on buildings, parking garages, or even church towers. However, for extensive digitization of urban spaces, mounting on streetlights is almost the only option.

SONAH has already completed over 80 IoT setups across Europe. A small selection of these projects can be found on sonah.world.

Our sensor complies with Protection Class II. This involves the use of double or reinforced insulation to reliably prevent electric shock.

Yes, the sensors are regularly updated by SONAH, the operator of the sensors. These updates ensure data and cybersecurity. The updates can be installed over the air, meaning no physical interaction with the hardware is required. Our customers can rest easy – we take care of everything!

Access to the data is determined by the client, often the cities themselves. The data is versatile and useful for different roles within the city. Live data can help drivers find available parking spaces, while historical data forms the basis for modern city and traffic planning for the relevant departments.

Ultimately, it is up to the client to decide who can access which data. We are happy to offer advice and support.

Our sensors require continuous power (230 V AC). Typically, a sensor consumes 10W.

1. Plan (streetlights / mounting locations + parking spots)

2. Perspective videos of the location (sensor angle to the parking spots)

3. Mounting location and type of installation

4. Mounting height

5. Is continuous power available?