CAM modules work with common IP cameras, i.e. digital native cameras, in the visible domain. They are the most diffuse and affordable cameras on the video surveillance market. So, no special cameras, like the expensive thermal ones, are required.

‍

Yes, CAM modules can be used with cameras from any manufacturers. SnowCAM, FloodCAM and SmoCAM are completely camera agnostic, while WeatherCAM requires few minimum requirements (but they are usually available on most of IP cameras).

• ‍Web API control: Cameras should provide WEB APIs for frame capture and parameter control (see Question #10 below).
• ‍RTSP: Cameras should provide real time video streaming for frame capture (if not available via Web API control).

‍

Rainfall intensity measurement requires, together with the images to be processed, information about the exposure parameters (ISO/gain, shutter speed, aperture, focal length) used to capture the images themselves. Two scenarios are possible:

• If you want to use the camera also for other surveillance tasks, you’d probably let the camera auto adjust its capture parameters, to always produce the most visible images in any lighting condition. In this case, WeatherCAM’s rainfall intensity measurement is available only for cameras that are able to dynamically provide, together with each saved image, the set of its related capture parameters.
• If you decide to dedicate the camera to rainfall measurement only, these parameters can be setto fixed values, so since they would be known in advance almost every camera could be suitable to work with WeatherCAM’s rainfall intensity measurement.

‍

CAM modules requirements for cameras’ hardware are quite easy to fulfill.

• Resolution: Ideal resolution is 1920x1080. Lower resolutions may result in less accurate result. Larger resolutions usually result in noiser pixels that may compromise overall CAM modules accuracy.
• ‍Sensor size: Sensor size should be 1/2.8” or above to provide a proper pixel size.‍
• IR illuminator/Ir cut filter: In order to allow nighttime / low light rain measurement and snow detection a IR illuminator is required (see Question #11 for more details). Cameras should be able to remove the IR cut filter in front of the image sensor.

Yes, CAM modules work with both bullet and PTZ cameras (dome or half-dome). On PTZ cameras WeatherCAM works with a fixed view on a preset camera position, and suspends its measurements when the camera is away from that position. Once the camera is back to its home position, WeatherCAM starts measuring again, automatically. Other CAM modules don’t require fixed views, so can work with any PTZ view.

WeatherCAM is currently available only for fixed cameras (for PTZ ones, see Question #5), while SnowCAM, FloodCAM and SmoCAM can also process images shot by cameras in motion, no matter if mounted on vehicles or drone. To return results in real time, cameras in motion should be able to run CAM modules onboard, or should be coupled with an edge processing unit. Otherwise, a stable internet connection to transfer captured images and perform server-side image processing is required.

‍

CAM modules are currently available only for server side execution. Servers can be on WaterView’s cloud infrastructure, based onAWS instances, or on customers’ own infrastructure. In both cases, images are transferred on servers to be processed, and data are returned to customer via API or dashboard. WeatherCAM can perform pre-processing tasks also onboard few selected camera models, to only transfer raw metadata on servers, thus limiting bandwidth occupation and reducing privacy issues. Compatible cameras are manufactured by Axis Corporation.

In 2021 WaterView is going to release stand alone versions of the CAM modules, able to run directly on cameras from major manufacturers (Axis, Bosch, Hanwha and AndroVideo mainly). Stay tuned!

‍

Cameras must be placed outdoor, on a wall or a pole, without being shaded by balconies, loggias or roofs. Ideal height of placement is 3-6 m from the ground, with the camera slightly pointed downward. CAM modules (especially WeatherCAM) don’t require cameras to be pointed toward the sky, and work at their best when cameras are pointed toward a contrasted background (like buildings, walls, hills or trees in the distance) preferably looking to the North. Including large water surfaces in the camera view, like the sea or lakes, should be avoided.  

CAM modules prefer images with wide views, where the main subject and its surrounding are both visible. Images too zoomed on the main subject are not suitable.

Cameras can communicate with local or remote/cloud servers in many ways, from fiber connection to 3G/4G wireless communication. Bandwidth occupation is mainly related to the dimension of the transferred images, but usually it’s no more than 18 MB/minute, that is necessary to upload 12 images at 1080p resolution per minute.

‍

SnowCAM, FloodCAM and SmoCAM don’t require specific settings. WeatherCAM’s Precipitation type analytics and rainfall intensity measurement require these parameters to be set in the following ranges:

• shutter speed:variable in the 1/1000 - 1/200 range in daylight, fixed at 1/100 during nighttime
• focal length(only for multifocal cameras): variable in the 3-5 range for 1/2.8” sensors

Moreover, snow detection works at its best when it’s coupled with temperature measurement based on onboard camera sensors (see Question #13).

‍

SnowCAM and FloodCAM can be performed during nighttime, if artificial illumination provided by street lamps or other visible light sources is available. IR illumination is not required.

SmoCAM is currently available only in daylight conditions.

WeatherCAM’ visibility estimation during nighttime requires street lamps or other visible lights to be included in the camera view. IR illumination is not required, visibility estimation works with both B&W and color images.

WeatherCAM’s precipitation type detection and rainfall intensity measurement require IR illumination and IR filter removal during nighttime or low light conditions.

To understand and decide what you can detect and measure with a single camera, you should first consider that, apart from visibility estimation that works with both color and black & white images, we have two families of applications:

• WeatherCAM’s Precipitation type monitoring and Rainfall intensity measurement require nighttime IR illumination and work with black & white images.
• Snow cover, flooded road and smoke detection work instead with color images, even during nighttime.

As a consequence, if you want to monitor all the phenomena that the CAM modules are able to detect, we usually suggest to devote half of your cameras to precipitation type monitoring and rainfall measuring, while using the other ones for monitoring road pavement condition, visibility estimation and smoke detection.

‍

Temperature measurements are camera model specific. The measures come from the embedded thermal sensors (cleaned from artificial effects due to camera functioning) and must be available via camera APIs or exposed in some other way, to be used by WeatherCAM’s temperature tracking feature.

Depending on the camera model, if not included in our compliance list, it may require 2 to 4 weeks to have a new model delivering reliable environmental temperature measurement derived from camera internal thermal sensors.

‍

No, CAM modules don’t need to be trained on each every new camera or location. When WeatherCAM starts monitoring a new location, it undergoes a 7 to 12 days self-tuning period to collect information about the light conditions of the camera view. These information are used by WeatherCAM to calibrate the physical model at its core, to return reliable precipitation detection and rainfall measurement data.

‍

CAM modules has been tested in real environment in comparison with traditional weather monitoring instruments and human observation. A technical document that summarizes the main results of the campaigns is available on demand.

‍

CAM modules cannot make a single camera as accurate as a dedicated weather station, like those connected to traditional Road Weather Systems. Nonetheless a video surveillance network equipped with CAM modules turns into an outstanding, non-dedicated Road Weather System that not only returns reliable data at the single camera scale, but let users have a better understanding of the overall weather conditions and their effects with an unprecedented spatial and temporal granularity. Being software only solutions, moreover, CAM modules have no recursive maintenance costs, and add value also to previous investments in video surveillance equipment.

‍

No, CAM modules simply transfer to the cloud and analyze snapshots (i.e. still frames) captured by the cameras, thus limiting a lot bandwidth occupation. To return more robust and reliable data, CAM modules analyze up to 12 images per minute, and return data with a time granularity of 1 minute.

‍

CAM modules usually are linked to the cameras via HTTP or RTSP connection. Images can also be uploaded to the cloud via FTP. For selected camera models, an image pre-processing module is provided to be installed onboard the cameras themselves; this software package not only pre-processes captured images, but also automatically establishes a HTTP connection with the server in the cloud. Camera names and locations (latitude and longitude) must be provided to correctly refer collected images and save processed data in the supporting geographical database.

‍

Yes, if you have camera models that support the execution of the local WeatherCAM pre-processing module, this software package is part of the SaaS offer (backend services based on Waterview cloud).

‍

Cloud services are the indispensable tools to interact and get the most from the data collected by your cameras. They come under the name of Atroona, a complete set of tools developed by WaterView to produce actionable weather insights for many businesses. Under a friendly graphic user interface, Atroona wraps up:

• an image processing engine powered by the CAM libraries
• a database to store raw weather data
• a complete set of tools to access, visualize and download weather analytics
• a notification centre to set automatic weather alerts based on events defined by the user
• administration tools to manage connected cameras
• a collection of REST APIs to smoothly integrate weather analytics into user’s workflows and/or third party solutions.

‍

The Atroona dashboard is accessible via web browser at the following URL: https://atroona.waterview.it/. Customers are provided with a username and a password to enter the platform and access, visualize and download their own data). Atroona lets you browse real time and historical data collected by each camera, plotting them in different time charts and letting you to download them in different file formats (Excel, CSV, pdf, etc). Atroona also offers a visual interface to create and manage notifications for all the weather events detected by your cameras.

‍

Yes, we also offer CAM modules and Atroona’s backend services also for on premise installations, especially for large customers (like national road authorities, electricity transmission networks, etc.) that have their own IT infrastructure and personnel, and prefer to internally manage data collected by their own video surveillance network. A service to create a backup copy of the database on Waterview cloud, however, is available on demand for on premise installation too.

‍

Atroona’s dashboard is not always necessary to use collected weather data. If you already have video management systems (VMS) or other infrastructure management tools that you use to centralize information coming from the field, Atroona’s APIs are the best choice to smoothly integrate data captured by the CAM modules into your workflow. The API reference guide is available on demand for subscribers of CAM products.

‍

Notification can be set for single or multiple cameras, and for each weather variable/event detectable by the CAM modules. Users can set their own variable/event thresholds (like “temperature has been lower than 2 °C for 30 minutes”, or “it’s snowing since 3pm”, and receive alerts via email or push notifications, to easily integrate communications into their workflows (i.e. open a pop-up message on a video wall). While the event/condition that triggered an alert is still ongoing, the alert is flagged as open, and no more alerts are sent until the event/condition terminates or returns under/above the set threshold. Then the alert is flagged as closed, and is ready to be sent once more if the event/condition happens again.

‍

CAM modules are tools to monitor weather events and their effects in real time, so they don’t return users with traditional weather forecasts. Nonetheless, depending on the number and placements of the cameras used with the CAM modules, it’s possible to track the spatial and temporal evolution of weather events and conditions, thus enabling nowcasting predictions like “the rain front is moving north, and can reach the city center in 1 hour” or “if the current precipitation and temperature conditions persist, the probability of ice presence on road pavement is very high”.

‍

The standard solution offered by WaterView to enhance a video surveillance network with weather monitoring features is represented by a subscription to the desired CAM modules, coupled with Atroona's backend services. Atroona and the CAM modules come in two flavours:

• ‍Software as a Service (standard offer), installed on WaterView cloud and served via web browser and APIs, for customers that do not need customization or prefer to rely on outsourced IT resources (annual subscriptions or upfront payment for multiple years are available).
• ‍On premises, installed on the IT infrastructures of customers that must rely on their own servers or require a high degree of customization (usually offered with one-off licenses).

‍

All the data captured by the CAM modules are anonymized, and are simply referred to the place where the camera is installed and the time they have been generated. Images are not stored in the cloud, unless the customers require for this possibility or for research purposes (only in pilot installations). In these cases, we offer a dynamic masking feature to blind in real time sensitive areas like faces or car plates.

‍