Neural Labs' solutions ecosystem for access control

Access control is one of Neural Labs' main sectors along with ITS, security, traffic, mobility, and logistics.

Inside access control area there are several scenarios and casuistries that will be treated during this article to make Neural Labs' proposal clearer.

Providing more and more vertical solutions is one of the company's objectives, without forgetting the clients that only need to buy "naked" technology.

Benefits

In general, the use of video analyses like license plate recognition and other additional characteristics like make, colour and vehicle classification have many benefits in the access control area. Some of them are: 

• Increased control over parking.
• Increased parking security.
• Access to parking automation.
• Decreased internal and external fraud.

Let's remember what applications the implementation of video analyses can have when accessing to a parking lot:

1. In case of a rotating parking lot (paid per minute), the license plate recognition system can simply transfer the data to the management system so that it runs the parking logic.   This system will usually link the license plate with the ticket picked by the user in order to, among other things, validate if ticket and license plate match when the vehicle exits.   2. In this type of parking lots, Neural Labs can also identify vehicle classification (car, motorcycle, etc.) and transfer the data so that the corresponding fee is applied.   In some countries, by law, there must be a different fee stablished according to the vehicle type. Detecting it from the entry can help to select the appropriate fee from start.    Historically, there has been other technologies used for this function, like vehicle loop detectors. Nowadays, with the irruption and easiness of deep learning, it exists this completely non-invasive alternative to determine the vehicle classification.   3. Automate the parking lot accesses (residential condominiums, companies, etc.) based on authorized lists.   The system would open the barriers if the license plate were on the authorized list(s). There can be several lists and include residents, employees, workers, providers, etc., each of them with their own schedule rules.

4. Managing suspicious vehicles lists or with a specific interest.

The system can handle surveillance lists and generates alerts if any of the vehicles in these lists is detected trying to access.

In shopping centres, vehicles that are known to be related with crimes or that once abandoned the parking lot without paying, can be added to these lists.

5. Independently audit your parking lot accesses.

Installed as an independent system from the parking management one. Some clients want independent access management to audit their existing parking management system, no matter if it's manual or semi-automated. Mismatches often appear from this audit due to internal and external fraud.

6. Verifying that the license plate is not cloned.

Neural Labs system can detect, in addition to the license plate: colour, make, and vehicle classification. This information, detected when the vehicle gets closer to the access, is contrasted with the vehicle information stored at the system.

If it doesn't match, the access is denied and generates the corresponding alert to the operator. In some countries the theft and duplication of license plates is common, and, for example, allowing the access to a residential condominium only with license plate, makes it vulnerable.

7.  Verifying the vehicle's condition at the parking lot entry.

Neural Labs system allows to add video evidence cameras as a support of the license plate recognition ones. Like this, the system stores video evidence of the vehicle's condition at the entry. This information is extremely useful to avoid false vehicle damage claims.

8. Parking lot occupancy control. 

By installing cameras at all entries and exits of the parking lot, the system can keep an occupancy control of the vehicles inside it.

In this case, the time spent inside the premises statistic is very interesting, and we can even generate alerts if any vehicles exceed a certain time of stay.

9. Automating access and increasing the traceability of ports and logistics centres.

Using intermodal containers IDs' recognition combined with the truck's license plate reading, the access to ports is mechanized, while being integrated to the logistics management system.

10. Wiegand card emulation.

Sometimes we want to integrate all the parking lot accesses with the presence management or people access software based on proximity cards that already is in the building or company. 

The objective is to translate the license plates that can access to the parking lot with the ID cards from the employees, residents, etc.

Through a Wiegand license plate converter, the access control system detects the license plate recognition cameras as proximity readers and the license plates as people's IDs. The access logics remains in the presence management software.

Technology implementation

In some countries it's unusual to find any kind of parking lot without license plate recognition systems. Even though this has been forced by laws in many ocassions, like in Spain, technology becomes essential to the business. 

On the other side, there's countries that are just starting to have these solutions. From Neural Labs we are confident that the benefits that these systems offer to mature markets soon will be known on those emerging markets, and, therefore, they will be deployed massively. 

Historically the cost of these solutions has been a barrier in developing countries. Most recently, with the low-cost smart cameras irruption and the embedded license plate recognition, this technology's implementation has become easier, although not always successful. 

From Neural Labs, we presented in March 2019 our Smart camera proposal Neural Edge for the scenarios where edge processing is requried and offering very high recognition rates.

Success Factors

Neural Labs always stands out in two key factors for this technology implementation success:

1. The first one is that there must be a previous analysis of the client's needs.

Recognizing license plates is not enough if it doesn't have an objective and, ultimately, doesn´t solve the client's issue. This issue can be the need of access automation, detecting an internal theft, etc.

In order to make the technology serve a purpose, it often has to be integrated to other systems like parking lot management systems, facial recognition ones, RFID, etc.

Thus, after the previous requirements analysis, it should be clear what the integration needs are and even the tailor-made developments that have to be done, to be able of meeting 100% of the specific parking lot needs.

2. The second success factor in good operation is the technology´s recognition rate.

Let's imagine that a barcode reader reads 70% of the codes. This would make any warehouse management application totally useless and futile. 

Same happens in this sector. The license plate recognition rate must be very high or it becomes totally unusable, unreliable and all the logic that it follows fails.

Most of the times a lack of recognition ends up on a manual correction, and that's why low recognition rates lead to premature abandonment of the system.

In summary, we always must answer these questions:

Why do I want to recognise the license plates?

How does it help my operations? 

What issue does it solve?

Which systems can work with the license plate reader?

And always demand the higher recognition rates.

Scenarios

There's countless scenarios inside the access control area. Neural Labs' proposal on each of them is different and with different verticality levels. Some of them are:

• Shopping centres
• Payment or free
• Access to companies
• Access to residential areas
• Rotation parking lots (paid per minute)
• Logistic areas or ports

Technology

Neural networks and, nowadays, deep learning are changing the world we live in by making the systems smarter. Although much remains to be done, current applications of these technologies are countless. Neural Labs has been using neural networks since the 90s and, in recent years, has deviated to deep learning for: 

• License plate recognition
• Vehicle make recognition
• Vehicle classification recognition (car, motorcycle, van, etc.)
• Vehicle colour recognition

 

Neural Labs' proposal

Neural Labs offers products to address the cases explained above. 

The first task the company always has is to listen to the client in order to detect how our technology can help its operations. After that, we design the solution that may involve some integrations or personalization to be sure that the system actually fulfils the objectives determined by the client.

Being hardware agnostic is on Neural Labs' DNA. That's why in our solution design we almost always include hardware that is easily accessible in any market.

The support and formation, including both the concept tests and the subsequent implementation, is also vital for the project's success and it´s a chapter the company takes special care of. 

In summary, there's much to be done in the access control area. Neural Labs bets on doing it with quality, commitment and with the objective of stablishing long-term relationships with its clients. 

Elías Valcárcel, Neural Labs CEO