Automatic license plate recognition (ALPR) solutions are becoming more and more popular among end customers who would like to identify vehicles and people before they enter a premise. Realizing their potential, many solution providers are offering ALPR systems now, making it difficult for end users to know what to select.
And it's not just the end customers who lack proper awareness. Many systems integrators (SI) are also not sure of the best practices while installing an ALPR system because of its relatively new entry into the market.
Speaking to asmag.com, Ranjith Parakkal, Founder and CEO of
Uncanny Vision Solutions, a company that offers AI-based ALPR and more in various markets including the U.S., explained the major factors to consider when selecting a solution, what to know during installation, and the common mistakes that need to be avoided.
Criteria for selecting an ALPR solution
The first step to getting an ALPR system up and running successfully is choosing the right solution. According to Parakkal, this process begins with getting the right surveillance cameras in place.
1. Matching the use case to camera specifications
A key factor to be considered for the ALPR system is the camera specification. Depending on the location of the installation, the camera needs to be more specialized. For instance, in places such as highways or high-density commercial complexes, the cameras should be sophisticated enough to ensure high-fidelity capture. Inevitably, this will increase the cost and the customer should be aware of that.
“A mismatch of specification in this regard would result in poor accuracy of the system due to issues with focus, motion blur, and poor illumination of plates,” Parakkal said. “For the same reason, it becomes important to validate the accuracy of the system at all times of the day before proceeding with the selection.”
2. Local processing versus processing on the cloud
A debate that’s often seen these days is the cloud vs on-premise processing of data. According to Parakkal, local processing would provide better results when it comes to ALPR.
“In many use-cases like authentication and integration with payments, it is desirable to enforce local processing, considering that the system needs to read the license plate real-time,” Parakkal said. “Processing on the cloud in such cases not only adds to the latency but also leaves the customer vulnerable to potential internet outages.”
3. Adaptability
The solution should be amendable when new regulations come in or change in license plate styles. In some countries in Asia, there is a lack of standardization in license plate design. In such situations, the solution should be flexible enough to detect different types of plates.
“As license plates evolve with time and changes to rules, it is important to pick a system that demonstrates adaptability and learns over time,” Parakkal noted.
What to know when installing ALPR
Once the best solution for your requirements is chosen, you should take a look at the best practices for its installation. According to Parakkal, this ranges from getting the camera configurations right for the environment to fine-tuning the reading latency levels.
1. Camera settings
Most ALPR systems need fine-tuning of the camera for day and night conditions. It is important that the SI configures the camera settings suitably to ensure human readable plates at all times of day, including night mode settings, exposure levels, etc. Besides, the position and height of the camera also play a crucial role.
2. Supplementary infrastructure
SIs should understand the volume of data that goes to external servers and provision the bandwidth accordingly. A good local network connection without bottleneck points would be necessary.
“In case of low-cost solutions, it is important to have vehicles come to near stand-still for best results, necessitating speed breakers or other such means to slow it down,” Parakkal said. “An additional issue pertains to physical obstacles like tailgating. Placement of the camera in the best position to avoid occluded views is highly desirable.”
3. Reading latency
It is important to keep reading latencies low while using it in authentication scenarios. Appropriate tuning of various camera and software parameters may help in achieving this.
Common mistakes ALPR customers make
Not getting the camera specifications right is the most common mistake seen in the field, according to Parakkal. For instance, when the shutter speed is not fine-tuned to match the speed of vehicles, the visuals can be unclear. Not getting the latent reading right is also a problem.
The quality of the ALPR solution, especially its accuracy levels, is of paramount importance. But many customers tend to not check such parameters and compare with others when selecting a solution.