Computer Vision and its application in Financial Services

Artificial intelligence (AI) has been infiltrating the finance industry over the last decade. Typically, due to strict regulation and data privacy barriers, industry incumbents have been resistant to change. However, as Fintech companies continue to grow in prominence, even the longest standing firms see the enormous benefits of AI innovations.

Banks and other financial services already use applications such as machine learning and natural language processing to enhance their operations. Increasing computing power, better connectivity, cloud platforms, and data on the edge (edge computing) are starting to take institutions to the next level. 

One of the most prominent applications today is computer vision, sometimes known as machine vision. The global computer vision market is expected to grow at a CAGR of 7.7% to reach USD 18.24 billion by 2025. In this article, we will take a closer look at what computer vision is and how it will transform financial services.


What is Computer Vision?

Computer vision technologies enable machines to visualize and analyze visual inputs like images and videos. The objective is not only for the machine to know what it sees, but to understand the context and take appropriate actions. Computer vision has had limited real-world applications for many years until the advent of deep learning models. The advancement in AI now provides the power to solve problems, with algorithms capable of accurate image analysis and feature extraction.

Moreover, the vast amount of visual data available online ensures enough data to train machines in real-world applications. For example, autonomous vehicles development relies on cars understanding the world around them. Using billions of images, the sensors in automobiles can put the environment in the right context and be more successful when operating without humans. 


The majority of computer vision applications will try to recognize things in photographs, and can broadly split into seven categories depending on the use case.

  1. Object Classification – provides the broad category of the object that appears in the image
  2. Object Identification – identifies the type of the given object in the image
  3. Object Verification – checks whether a specified object appears in an image
  4. Object Detection – finds where the objects are in the image
  5. Object Landmark Detection – looks for the key points of an object in an image
  6. Object Segmentation – works out which pixels belong to the object within an image
  7. Object Recognition – discovers which objects appear in the image and shows where they are

In the example below, you can see how computer vision systems break down images to come to a decision.



Applications of Computer Vision in Financial Services

There are several ways in which computer vision is being deployed in financial services.


Document Extraction

Financial firms are often swamped with large numbers of incoming paper documents and forms typically administered by human workers. Companies such as HyperScience work with institutions to automate document data extraction with machine vision.

The software digitizes paper-based information before segregating it for human review. Traditionally, the process is done manually and is highly error-prone as well as being time-consuming. Image recognition technology integrates with existing processes, automatically identifying and understanding the documents.

Through automating document extraction, teams can focus on solving problems, rather than data entry tasks. In the case of complex contracts, computer vision algorithms can quickly find the salient points, mitigating the need for humans to study them. 

Captricity offers a similar solution and helped New York Life Direct to automate data digitization. The solution sorts through over 100 different versions of their insurance promotions and integrates with legacy systems. The result is a reduction of 50% in total operations cost by decreasing processing time from days to hours. 


Know Your Customer (KYC)

The Spanish bank, BBVA, uses face recognition to confirm the identity of customers when submitting a photo ID. During a video call, customers can open an account using their smartphone, eliminating the need for manual verification. The KYC process is reduced from hours to minutes, thanks to computer vision technology.

KYC processes tend to require much documentation. Extraction processes using machine vision can speed this up when there is a requirement and enhance mortgage, loan, and credit card applications. Banks get fewer errors, and customers receive a much-improved experience.


Supporting a Cashless Society

Computer vision is helping financial firms as consumers continue to favor digital payment channels. In 2017, Wells Fargo officially launched cardless technology at all of its 13,000 ATMs. Customers can log into their mobile app and receive an eight-digit code that gives that use to all ATMs in the country, rather than carrying a physical debit card. 

Customers can initiate a cardless ATM transaction by opening a mobile wallet on their smartphone and holding it near an NFC-enabled terminal at an ATM.

The major tech companies like Apple and Google already have mobile payment systems that can remove the need for carrying a physical card. Point of sale biometrics allow transactions against digital details which is likely to be the future of banking.


Claims Processing and Underwriting

Ant Financial is a Chinese fintech vendor with over 7,000 employees. Their app can recognize damage to motor vehicles and facilitate the claims process using computer vision.

The app will analyze an image, understand the severity, and send a report to human insurance agents. It can recognize the parts and how they should be repaired, as well as how the incident will impact the premium of the driver. Ant Financial train their model using thousands of images of damaged cars, which are labeled against the cost to fix them. Pictures are stored with different lighting and angles so that the app is accurate and always learning. 

Cape Analytics computer vision platform can analyze satellite imagery for the property insurance market. The4 software will find attributes that the insurer will be interested in and report back on potential damage or quality issues. For example, it will show if a roof is prone to damage on a property. Insurers can use Cape Analytics to improve the underwriting process to make a more informed decision about the risk of a property.


Reducing Fraud

Computer vision techniques like facial recognition and retina scanning can help financial institutions to improve security procedures and reduce the risk of fraud. Uncanny Vision specializes in providing intelligent surveillance cameras for banks and ATMs.  The technology will detect any suspicious activities and proactively alert authorities, preventing malicious behaviors.

Several banks use biometric security within mobile apps to improve user authentication. Wells Fargo, Bank of America, and Barclays all make use of fingerprint, retina, and facial scanning so that if a phone falls into the wrong hands, it is still safe from criminals.


Capital Markets

With a vast amount of geospatial data becoming available in recent years, computer vision technology can track activity at very granular levels. Investors and economists can quickly see US retail traffic, automotive activity, and parking lot usage. Hedge funds can use such data to understand retail patterns better and get ahead of the market.


Challenges for Computer Vision in Finance

Despite the benefits to banks, consumers are still cautious about new technology, especially when it involves financial transactions. There are still plenty of people who would instead visit a branch and speak to a human than rely on their smartphone for making a payment. Financial firms need to help educate their customers on using digital services, without compromising the security of personal information.

A further barrier could be the cost of the technology. Time, effort, and capital are needed to bring in the right software and hardware, as well as to train it with data. Most financial institutions don’t have the resources, whether in cash flow or skilled staff, to create the relevant computer vision models. Most firms will have to rely on external vendors to build their computer vision applications, rather than hosting them in-house, raising additional data security and privacy concerns.

In the case studies noted in this article, computer vision relies upon vast volumes of labeled images to be accurate and make the right decisions. For example, you can’t merely buy a platform and start feeding int new documents without training it on what to look for. To get to the right level can take years of experience for machines unless enough goes into developing them.



Computer vision technology is beginning to have a significant impact on the financial services industry. As the digital ecosystem continues to grow, firms have a wealth of opportunities to deploy computer vision as part of their ongoing innovation strategy. The need to enhance processes, reduce costs, improve customer experiences, and meet demand is imperative as the fintech competition looms large.


Disclaimer:  The author of this text, Jean Chalopin, is a global business leader with a background encompassing banking, biotech, and entertainment.  Mr. Chalopin is Chairman of Deltec International Group,

The co-author of this text, Robin Trehan, has a Bachelor’s degree in Economics, a Master’s in International Business and Finance, and an MBA in Electronic Business.  Mr. Trehan is a Senior VP at Deltec International Group,

The views, thoughts, and opinions expressed in this text are solely the views of the authors, and do not necessarily reflect those of Deltec International Group, its subsidiaries, and/or its employees.