Solutions may be brand-agnostic, but camera quality still matters.
Most vendors of video analytic solutions for traffic management will tell you that the camera brand doesn't matter to them as long as it can capture footage that meets the minimum requirements. The minimum requirement is that the object that you want the system to identify should be identifiable to the human eye. In other words, if you can't see it in the video, the computer can't either.
Camera resolution
A general rule of the thumb is that the higher the resolution, the easier it is for the computer to see things in the footage. However, many video analytic solutions can even run on age-old VGA cameras, as long as it can capture visible video. According to Daniel Stofan, CEO of GoodVision, his company’s solutions deliver the best performance at 1280px x 720px and 1920px x 1080px (FULL HD) resolutions.
"Generally, GoodVision can handle lower resolutions all way down to VGA," Stofan wrote in a post that was published in Medium. "However, lower resolutions go hand in hand with low-quality optics and low bitrate, causing the object contours to be not crisp (blurry) or not resemble the object from the real world. Set the resolution which displays the object's contours clearly."
Framerate and shutter speed
Setting your camera to the right framerate is essential for accurate capture of an object's motion in the video. For video analytic solutions, this is important for the proper functioning of the tracking feature.
According to Stofan, tracking requires solid object trajectories, i.e., for the origin-destination counting of traffic. Moreover, the higher the speed of objects in the video, the worse it affects tracking if the framerate is low.
"Ideal FPS [frames per second] for video-analytics which works well with most of the scenarios is between 10 and 30 frames per second," Stofan added. "The bigger, the better, however, FPS bigger than 30 per second does not have any visible impact to tracking quality. Lower than 10 frames per second cause tracking problems, especially in crowded scenes and with fast-moving objects, which are literally "jumping" from place to place over the scene."
Edge and server-side
Analytics for traffic and road safety management has to ideally be a combination of edge-based and server-side analytics, according to Ranjith Parakkal, CEO of
Uncanny Vision. This is because some of the features that require immediate response need to work on the edge.
"Access control, for example, has to be edge-based," Parakkal said. "This is because the access control system needs functions in real-time, to provide or deny access to a person as quickly as possible. For instance, if a vehicle arrives at a boom barrier, its license plate is read, cross-checked with a database, verified, and the access is granted. All of this has to happen within seconds, and for this, the analytics has to be on edge."
Other applications may be a mix of edge-based and server-based analytics. For instance, identifying an object can be on edge, but reading its license plate can be on the server. Having said that, Parakkal added that they try to keep everything on edge to avoid any complications that may arise due to network problems.
Conclusion
To sum up, many vendors prefer to keep their solutions camera brand-agnostic. But the resolution of the video and shutter speed does have an important role to play in the final output.
Quality of data is critical to getting the best results from your video surveillance data. One of the major concerns, especially in many public projects, is that the data that reaches the server may not be complete. This results in inaccurate video analytic readings.