In a recent Benchmark group test looking at object-based video analytics, the RC3502HD-5211 camera from Riva was included. This device has embedded video analytics from VCA Technology, and includes what is referred to as object-based detection. During the test it became apparent that whilst the embedded analytics worked well, they weren’t best suited to tracking objects which were left or removed from a viewed scene by an individual. However, they did offer a wealth of other options which Benchmark further explores here.
[dropcap]T[/dropcap]here can be no doubt that video analytics have become an everyday option for many applications. Whilst there are systems that deliver very complex solutions, the choices from embedded IVA or app-based systems make the technology very appealing.
The deployment of IVA starts with a good understanding of what the system is setting out to achieve. For example, line crossing analytics are straightforward when it comes to having an understanding of what to expect. A virtual line is created, a target size defined, and if relevant directional discriminations are applied. When a target crosses the line in the prescribed direction, an event is reported.
Object-based analytics are not so simple to define. This is because the term ‘object’ can cover a multitude of targets including – but not limited to – people, vehicles, containers, briefcases and small assets.
So what does ‘object appear’ mean? Is it a car being driven into a detection zone, or something appearing without having been driven or carried into the detection area? Also, ‘object disappear’ could detect something leaving a detection area, or will the algorithm only detect items that disappear without leaving the zone?
During our recent Benchmark group test of object appear/disappear IVA, we looked at items being left in protected area by people passing through, along with items being removed from that area, again by people in the scene.
The Riva RC3502HD-5211 was included in the test. This camera features embedded video analytics from VCA, and the supported rules include object detection. However, these are more suited to objects which appear or disappear from the viewed scene, rather than detecting items which are abandoned or removed by an individual.
Because of this, the camera did not fulfil our test criteria, and despite having a good and flexible range of analytics rules, it could not be recommended in that specific test. Therefore, we’re taking a look at the VCA analytics independently to get a better view of performance across the board.
The RC3502HD-5211 is an HD1080p static dome camera equipped as standard with core intelligent video analysis functionality. The IVA, powered by VCA Technology’s analytics engine, includes intelligent video motion detection and tamper protection as standard.
Further VCA filters are optionally available at an extra cost. These include intrusion detection, people counting and object detection packages, and allow application specific scenarios to be created.
The camera offers H.264 and Motion-JPEG compression, and features include dual streaming, edge recording, audio support and digital WDR.
The camera uses a 1/2.7 inch CMOS sensor to deliver HD1080p and HD720p streams, plus other standard definition resolutions, at frame rates of up to 25ips. It is important to note that if an HD1080p stream is used while the VCA functionality is enabled then the maximum frame rate drops to 15ips. With a HD720p resolution, real-time streaming can be preserved. Maximum bit-rate is 6Mbps.
The dome unit is equipped with a 6-9mm varifocal lens, two way audio and alarm input/output. The typical video configurations – BLC, DNR, AGC, sens-up, shutter control, privacy masking and AWB – are all present. Other functionality includes electronic image stabilisation and defogging.
Power is PoE or 12V DC. The unit features an integral heater, but this only functions if 12V DC power is used.
The physical installation of the camera throws up the first of a couple of niggles. The camera housing has two cable entry points: one on the side of the unit and the second on the rear. The latter is the one that most installers and integrators will use.
A Cat 6 cable with standard RJ45 boot is a very tight fit, to the point where you’ll find yourself physically forcing the plug into the socket because it is located so close to the housing. Both connecting and disconnecting the camera leave you with a feeling that some type of damage could occur.
As the dome powers up another small niggle raises its head. The camera’s fan unit cooling the main module constantly emits an audible whirring sound. If anything, it sounds very much as if it is overworking. On a few occasions it fell silent, but it was only for a few seconds. Having dealt with cameras from Riva before we can’t remember a similar issue, so it does seem to be specific to this model.
Once the camera is connected and powered up, it’s time to start the configuration process. The camera is provided with a simple IP utility and this finds the camera almost immediately and allows the network configurations to be made. The one thing it lacks is a way of opening the camera’s web page; this needs to be done manually.
With the network settings completed, you can then access the camera’s internal web pages. The viewing elements load automatically and final configurations can be carried out.
The camera is supplied with a USB stick which includes the IP utility and two PDF manuals. One is for the camera and the other covers the video analytics.
The menus are straightforward and intuitive. They are split into Basic Configuration, Video and Audio, Event Configuration, Network Configuration, VCA, Peripheral, Maintenance, Activation and About. General camera set-up is well catered for, and allows a high degree of flexibility.
The VCA menu covers the camera’s integral analytics set-up. While the interface might at first appear to be a bit basic, it does actually work well, and is simple to follow both during initial set-up and when making any subsequent tweaks to the analytics configurations.
Riva’s cameras offer a range of analytics rules; all but the most basic require a license. However, the breadth of available rules does make the investment worthwhile. Also, when setting the camera up, the menu structure is easy to follow once you have a handle on how the analytics work.
There are a number of rules and filters: presence, enter, exit, appear, disappear, stop, dwell, direction, tailgating, speed, colour and count line. Objects – basically targets – can be defined as people, vehicles or ‘clutter’. The latter is basically a catch-all term for busy scenes. There is additionally an option to define ‘unclassified’ objects. This allows the setting of minimum and maximum sizes for custom classifications.
It is worth spending a bit of time working through the VCA options before installing a camera on a live site. It’s all relatively straightforward, but you will need to have a feel for how the zones work. Also, before you start, make sure that analytics is enabled. This requires a small box to be ticked, and it’s easy to miss!
The menu order is slightly out of sync, with the creation of rules coming before calibration of the IVA. Calibration is simple, requiring mounting height and tilt angle. It makes use of visual ‘mimics’ to allow person sizing. This allows fine adjustment if you’ve estimated the angle of the view.
With calibration completed, the option to create a new line, zone or counter is accessed via a contextual menu. This also gives options with regard to display.
The set-up of the actual rules is straightforward, and double-clicking on the Zone listing reveals additional parameters. These include the exact type of analytic to be deployed. Multiple choices can be selected if you are seeking to optimise the IVA performance.
The enter/exit and appear/disappear rules have similarities, in that they detect a classified object in a detection zone, but there is an important difference. Enter/exit rules look for objects that are detected entering or exiting the zone, while appear/disappear looks for objects that do not travel into or out of the protected area. Instead it looks for objects appearing and disappearing, such as coming out of a doorway or passing through an exit, inside the protected area.
All of the analytics offered accurate and consistent detection, within the parameters of the rule definitions. We did experience a few nuisance activations, but these were simple to deal with. Because zones can be configured with numerous nodes it wasn’t difficult to work around problematic areas in the images. Alternatively, non-detection zones can be used as masks without preventing the area from being viewed.
Performance remained high during varied weather conditions, including rain, hail and the odd snow flurry. Night performance was also good, and under IR light detection retained accuracy. No genuine events were missed, and nuisance detections were minimal.
The VCA also has options for tamper detection and global scene change, and both work well.
The VCA analytics also allow the use of logical rules. The definitions are quite flexible, ranging from a single rule violation through to multiple triggers (and/or options). There are also options to specify time (duration, time window, date or even elapsed time since an event), input/output or count values.
This adds a significant value to the analytics, and delivers benefits for installers and integrators creating value-added solutions.
Riva’s cameras are decent units, and won’t disappoint with regards to performance, However, the main reason for specifying them will be the VCA analytics, which add significant value to the proposition!