THE CHALLENGE
Develop an advanced solution that supports and facilitates the recognition of profiles of different types of keys, identifying the correct Silca key.
THE SOLUTION
A software based on an Artificial Intelligence algorithm, trained to correctly recognize different types of keys through a Synthetic Training Dataset.
Silca S.p.A | Key Duplication
Silca S.p.A is a global leader in key duplication services, offering innovative and technologically advanced solutions dedicated to the security sector. In its latest ambitious project, Silca collaborated with Uqido to develop Artificial Intelligence software to identify key profiles.
Silca Profile Reader is a tool that makes identification — the most important and complex part of key duplication —fast and advanced.
IMPACT
Silca Profile Reader is a completely innovative solution that, through Artificial Intelligence algorithms , facilitates the process of identifying key profiles and simplifies the duplication activity for new operators, speeding up the work of specialists.
Needs
Silca needed to simplify the key duplication process, traditionally performed manually. To make a copy, the locksmith must determine the profile of the blank key, choosing from thousands of available options.
Silca aimed to automate the recognition of key profiles, simplifying the process and reducing the margin of error. Therefore, we decided to develop an algorithm that, through Computer Vision, could analyze a photo of the key to be duplicated and recognize the corresponding profile.
Development | Synthetic Dataset Engine
To train the algorithm to recognize images, we needed photographic samples of the objects to be identified. However, in Silca’s case, manually reproducing all the photos for this training dataset was impossible due to the vast number of existing Silca key profiles. In agreement with the company, we developed Artificial Intelligence software capable of creating synthetic images, which are samples of the keys for training purposes.
Thus, the Synthetic Dataset Engine (SDE) was born. This platform is useful for anyone who needs to classify objects using Machine Learning technologies without having a real dataset. The resulting images are not photographs but high-resolution digital reproductions, created from the CAD models of the original keys.
Development | From SDE to Profile Reader
Thanks to SDE, we could give Silca a precise and functional algorithm integrated into the company’s app through the Profile Reader feature. By accessing the mobile app, the locksmith selects Profile Reader, takes a few photos of the key to duplicate, and receives a clear response on which model to use for the copy.
Profile Reader identifies keys available in the market, with a database of nearly 6,000 profiles continually expanding and often very similar, ensuring fast and accurate service.
“Today, the whole world is talking about Artificial Intelligence. Some companies have already had the courage and ambition to invest concretely in their future by establishing the necessary processes to adopt and leverage AI to their advantage. Silca’s team embraced this challenge years ago and can now boast a groundbreaking innovation for its market, reaffirming its leadership in this sector.”
Pier Mattia Avesani -Uqido CEO
Advantages | Why companies should choose SDE?
Reduction of time and costs associated with creating training datasets: SDE addresses the biggest challenge faced when utilizing Machine Learning technologies—creating a sufficiently large sample of data to train the algorithm effectively within reasonable timeframes and investments.
Quality control: With SDE, it’s also possible to execute and automate quality control processes, enhancing operational efficiency and reducing human error.
Versatility: SDE is a universal tool applicable to any classification and a wide range of needs.
The platform can support various types of activities, such as:
- Defect recognition on a product and quality control
- Recognition of specific objects
- Recognition of text or logos
- Recognition of jingles or phrases in video and audio
- Classification of volumes, quantities, or dimensions