With the accuracy of 96% and processing speed less than 1 second, VinOCR is being applied in solving problems related to remote customer identification. It is especially helpful during the Covid-19 pandemic as reducing social exposure, also maintaining and improving business efficiency and user experience.
Customer identification is the process by which businesses, especially financial institutions, collect and verify information about customers’ identities and addresses. With traditional methods, service users must go directly to the headquarters so that employees can visually observe, authenticate identity cards/citizenship, compare facial recognition features with data on records and papers.
Because of the cumbersome procedure in the authentication process, the service user’s experience is significantly reduced, even in some cases customers abandon the brand. For businesses, traditional customer identification also causes financial and human resources burdens, affecting production and business efficiency. In the context of the Covid-19 pandemic, the limitations of this approach are more evident than ever.
VinOCR quickly and accurately authenticates various types of documents
Recognizing the difficulties of enterprises in customer identification, Vingroup Big Data Institute (VinBigdata) has invested in researching VinOCR, belonging to the VinBase Vision product line. It applies computer vision technology to support written information detection, ientification and extraction from images.
VinOCR are developed on the basis of the advanced technology that VinBigdata owns, in which the outstanding highlights are: 93% accuracy in business card reader, extraction, classification, and directory storage of data, outperforming similar products in the market and up to 97% accuracy in character recognition for Vietnamese and English documents, which is currently the highest in the market.
With the superior technological capabilities mentioned above, VinOCR is capable of processing both formatted and non-formatted documents, tables and graphs, also handwriting. Taking less than 1 second, the software can read, identify and verify identity cards, passports, driver license, vehicle registration, household registration book and birth certificate. The accuracy of this task is up to 96%.
VinOCR promotes digital transformation of business operations
No more wasting time and resources for manual data input, with VinOCR, customer identification information is extracted and digitized automatically in real time. Service users, therefore, will have a better experience, thanks to significantly shortened filling time. At the same time, businesses will save a lot of time and effort for data entry, as well as minimize the risk of errors for input data.
This is considered one of the prerequisite steps of the roadmap for business digital transformation. The solution can be easily customized, integrated and conveniently deployed into the enterprise’s system. Thus, VinOCR has been applied in a number of businesses, including the VinID e-wallet. In the context of a strong digital transformation process, deploying VinOCR means that businesses will gain a significant advantage in the market.
VinOCR – A simple customer identification solution during the pandemic
During the Covid-19 pandemic, VinOCR is the solution to both ensuring social distancing and improving service quality, optimizing production and business efficiency. Data processing steps of VinOCR include:
- Step 1: Identify the type of image to support.
- Step 2: Image analysis, including text detection; text reading and content analysis.
- Step 3: Convert images to text.
On the client side, with the support of VinOCR, the authentication process is also shortened and simplified, just by uploading two-sided photos of 1 of 2 types of identification documents or the front of the passport, then waiting for the system to read and process it.
Electronic know your customer (eKYC) is one of the applications of VinOCR. In addition, products can be developed to solve many different problems, meeting various needs of businesses, such as automatically reading business cards, digitizing documents, etc.
Based on artificial intelligence, VinCamAI can analyze different facial attributes from age, gender to emotion. It is a technology solution that helps businesses optimize image dataset, providing a basis for decisions making.
According to the MarketsandMarkets report, the size of the global image data analytics market will reach $ 6.51 billion by 2022, with a compound annual growth rate (CAGR) of 20.4%. The main driver of this trend is the rapid growth in both quantity and diversity of business data (including image data), as well as the need for data-driven decision-making.
In particular for retail businesses such as convenience stores, supermarkets, commercial centers, etc, customer image data now becomes a precious resource, serving advertising campaigns effectivity statistics, target customers identifying, customer behavior analysis. Thus, not only for security purposes, these data also create direct impacts on customer shopping behaviors and business performance.
VinCamAI analyzes customer facial attributes
Compared to traditional footfall products, artificial intelligence (AI)-based solutions are capable of collecting and analyzing a larger amount of data, thus performing more tasks in the same amount of time.
The strongest evidence is VinCamAI, a smart camera developed by Vingroup Big Data Institute (VinBigdata). Applying computer vision technology, besides counting people, the product can also identify customer information (such as gender, age, emotion), identify VIP guests, foreign visitors and report strangers. The accuracy of face recognition and analysis technology reaches 99%.
With this feature, from the collected image data, businesses easily build portraits of the target customers, conduct in-depth analysis and make appropriate changes. This is also the basis for personalized marketing campaigns, helping recommend products to the target audience, thereby increasing the chances of being selected.
VinCamAI analyzes customer behaviors
Not only statistics of the total number of customers at the supermarket (hourly/daily), VinCamAI also supports behaviour recognition, and movement tracking. The data about crowded areas or shopping time of each area are also recorded by the system to determine the location that attracts customers, serving the distribution of products and resources support.
Combined with facial recognition and customer information analysis, the collection of behavioral data helps VinCamAI make insights about the shopping trends of each specific customer, for example: young customers are often interested in fashion boutiques, while older groups spend more time stopping at food stalls or promotions. The information will be a suggestion for businesses to set up customer-oriented changes.
VinCamAI speeds up payments
Another activity of retail supermarkets is payments. In fact, the speed of the payment process greatly affects the level of customer satisfaction and the assessment of service quality. With VinCamAI, the system will quickly determine the number of people waiting for payment as well as the amount of time for this task. This is the first step for retail businesses to identify and accurately detect the limitations of the service, aiming to gradually build a database as a premise for making important decisions.
Thus, with the new solution supported by AI – VinCamAI smart camera, supermarkets, commercial centers, and retail businesses will have a more powerful assistant in collecting, processing and utilizing image data of customers, from which to specific applications to optimize production and business efficiency and enhance user experience. In fact, VinCamAI products have been deployed in some Vincom Retail shopping centers, giving statistical results of 30% better than the traditional counting system (Footfall).
VinCamAI is one of VinBase Vision products developed by VinBigdata. Besides VinCamAI, currently in this product line, there is also VinOCR – a system that supports written information detection, ientification and extraction from images.
