Author: Ethan Ace, Published on Dec 19, 2017
Axis is expanding into a new market – radar.
The company has released their D2050-VE Network Radar Detector, claiming to minimize false alarms, in any weather, day or night.
We bought and tested the D2050-VE to see how it performed, testing:
- Detection range
- False alarm issues
- VMS integration
- Setup and calibration
- PTZ autotracking
And more. See our full results inside.
Based on our tests of the Axis Network Radar D2050-VE, there are several advantages:
- Solid detection across the specified range: The D2050 reliably detected humans and vehicles with few issues (discussed below) at distances up to 50m, its specified range.
- Few false alarms: Small animals, limbs, foliage, debris, and other common sources of false alerts in other detectors did not trigger the D2050. The only false alarms seen during testing were due to the pole it was mounted to swaying in high wind, eliminated by filtering swaying objects.
- Easy VMS integration: The D2050 added as an Axis camera to all VMSes tested, displaying a map overlayed with target location, reducing integration required compared to others, which generally send object location metadata and require the VMS to display this information on a map.
- Simple setup and calibration: Setup of the D2050 was simple, requiring only a reference map to be added and calibrated, along with mounting height, taking only a few minutes to complete.
However, users should beware of these key limitations:
- Metal objects limit detection: Moving between or behind metal objects such as cars and trucks causes spotty detection, or at worse, total misses of human subjects. Users should carefully consider placement of the radar as rows of vehicles may cut off detection.
- Limited PTZ autotracking: Despite Axis’ claims that the D2050 “Enables autotracking for PTZ cameras”, auto tracking was limited and could only be set to low zoom levels/wide FOVs. Increasing zoom level resulted in subjects being partially cut off or fully out of the camera’s field of view.
Note that Axis claims little to no impact on radar performance due to weather. However, we have not yet tested this as conditions have been fair. We plan to test in rain and/or snow as available.
The Axis D2050-VE sells for ~$1,200-1,300 USD online. A VMS license will typically beneeded for each unit for monitoring.
Pros Vs Analytics
Compared to many video analytics on conventional cameras, the D2050-VE performs better, completely ignoring common false alarm sources such as shadows, lights, wind blowing foliage or debris, etc. These issues cause some false alerts in even the top analytics we have tested.
Compared to video analytics on thermal cameras, false alerts will generally be much closer but thermal cameras are generally expensive. One clear plus of thermal plus analytics is long range, given the Axis radar short distance limit.
To the extent that one finds robust enough video analytics (either current offerings or upcoming deep learning ones), they will likely be more cost-effective than the Axis radar. Once you add in a separate camera, the cost of Axis radar itself, the cost of installing the radar and the cost of the VMS license, that is a significant amount of money if a ‘smart’ camera can do video and alerting effectively in one.
The D2050 is a wall mount unit about 12″ wide by 8″ tall, with the bottom (black) section essentially used for cable entry and the radar module contained in the upper, rounded portion.
The radar’s mounting plate is flat, but the radar module itself may be angled at +/- 25°, enough to move the coverage area to the side of a building, for example. However, no articulating mounts are available, so the radar may not be angled to cover odd-shaped areas which are not best covered with the 50° mounting range, nor aimed up or down to compensate for slopes.
The D2050 includes one Form A relay and 4 configurable I/Os which may be used to trigger other devices on specific radar motion zones. Note that only the Form A relay may carry voltage.
We review these features in this video:
Reference Map Calibration
Calibrating the radar requires only a few steps, asking the user to input mounting height and pick two points on the map (shown below). In our tests, detection distances were accurate after this brief setup.
RMD Zones Instead Of VMD
Instead of VMD setup found on IP cameras, the D2050 includes configuration for Radar Motion Detection zones (RMD). Users may create up to 12 include or exclude zones which may activate separately, allowing one zone to cover only a parking lot while another covers the field beyond, for example.
Zones may be set to trigger on human sized objects and/or vehicles and/or small and unknown objects to find tune detection.
The D2050-VE adds to VMSes as a camera, instead of using other integrations or plugins, and uses a VMS camera channel license. This reduces the amount of integration work which must be performed, as it uses the Axis VAPIX API. However, this means sites with multiple detectors must view them all separately, instead of being able to place them all on a single map view. Other perimeter detection options (SpotterRF, Optex, Southwest Microwave) typically send events and location metadata, not a video stream, and require additional VMS integration to place these objects on a map view.
Milestone and Genetec both integrate radar zone events, which may be used to drive alarms, bookmarks, etc. The example below shows separate bookmarks for two radar zones in Milestone:
In our tests, the radar detected subjects reliably out to ~50m/~164′, its specified range, true of both vehicles and human targets. Performance was slightly better when tracking objects moving across or away from the radar instead of approaching, but range was nearly the same in both cases. In the example below, our subject walks
Additionally, we saw no false alerts caused by blowing objects or small animals, despite the radar’s coverage extending into the brush shown above.
Metal Objects Block Detection
Metal objects such as vehicles, outbuildings, large electrical poles, etc., degrade detection performance or prevent it entirely. For example, when the subject moves behind the vehicle in the clip below, the radar fails to track him until he is past the other side. Users should beware of this issue when placing the radar, as lines of vehicles or even small outbuildings/sheds may significantly degrade detection.
PTZ Autotracking Poor
The D2050-VE includes a PTZ autotracking ACAP package installed from the factory, used to direct Q61 series PTZs to radar detected objects. In our tests, this tracking did not function properly with default settings, as high zoom levels cut off subjects or missed them entirely. However, autotracking works with lower zoom levels (~10-20% instead of 50% default), but fine details may be unavailable.
The example below shows a subject starting at ~125′ and walking near and past the radar and camera, automatically tracked properly throughout.
Axis says they plan to improve autotracking in future firmware releases.
Firmware Version Tested
The most recent firmware available at the time of testing was used: 18.104.22.168