What if computers had eyes like us and they could understand what they saw? That is the bet behind AI-powered computer vision. While streaming video from a set of cameras to a computer is already decades old technology, helping the machine to be aware of
what is in sight and take decisions based on the live information is still something scientists are working on to get it perfect.
Until now, computer vision has been used for quality control in industrial processes like welding, painting and other jobs performed by industrial robots. It is also the mastermind behind package sorting in postal hubs or airports. But these are all repetitive,
usually code-based tasks and don’t consider the interaction with real people or the needs of potential customers.
Computer Vision in Financial Services
Although it does sound far-fetched in the beginning, computer vision can make a real impact on financial services as we know them. It will disrupt and transform current practices only to replace them with a new paradigm. Let’s walk through the four primary
fields: banking, insurance, capital market, and commerce to assess the impact on every level.
The banking industry is a slow adopter of new technology, primarily due to privacy and security regulations. Computer vision will have a hard time penetrating this fortress, but eventually, it could prove useful for the “Know Your Customer” process which
right now is more of a problem of database joining than dealing with a real person. Using your phone to open an account seems farfetched, but if you think that your iris is more personal than your handwritten signature and unforgeable, it becomes a better
Another area in which computer vision could work for banks is retiring cards and replacing them with much safer single-use codes generated through apps installed on clients’ phones. Wells Fargo has already replaced cards with codes and more changes are on
As the technology evolves, it will penetrate more into our daily operations and become the norm. The winning philosophy behind this is that computer vision will eliminate a lot of friction and pain points related to identity verification and will eliminate
a lot of fraud risks.
Learning from the current application of computer vision in quality control, the insurance business could benefit from the technology by adapting it to its own needs regarding the accuracy of damage evaluation. Right now, this task is carried out by human
inspectors, trained in identifying the situations under which a loss occurred and assessing if this was an unfortunate case or if it was done with evil intention, to collect the policy.
All this work will soon be replaced by an app from the insurance company which could be used directly by the insured party to scan damaged goods. The images would be automatically compared with those from the database and a decision made on the spot, making
the inspector obsolete.
Another use in this area is improving the underwriting process dramatically. Right now, it is a combination of past manual steps and some digital components. Soon, we can expect an entirely digital approach, with improved UX and better pricing strategies.
Assessing granular risk will make the difference between top insurance companies and laggards. As lower risk clients will be offered better rates by those companies who adhere to the new technologies, other companies will be left with sub-par clients with
Computers armed with the gift of vision can make a real difference when it comes to stock trading. Although it seems like something out of a spy movie, being able to monitor traffic, up to a granular retailer level could show which areas are developing and
trigger an investment signal. For example, if there are increased traffic congestions in the regions which previously did not have such problems, that is a good proxy for increased economic activity, signaling a developing area.
Such data is essential for hedge funds, venture capitalists and other types of investors who are always looking for the next big thing.
Computer vision will also be adopted by retail chains since it has the potential to cut some serious costs incurred by staff. Imagine the savings a supermarket could have if they didn’t need cashiers, security staff and merchandisers.
This is already being tested in the AmazonGo concept store. The customer walks into the store, gets whatever they need and simply walks out, and is charged the correct amount on their debit or credit card. How is this possible?
The CCTV system in the store records all movements, a computer identifies each product and the action of putting it into the basket or removing it from the basket. For higher accuracy, this could be combined with RFID tags.
The applications don’t stop at unmanned stores. Taking some inspiration from quality control, the computer vision will be able to identify fake products by looking into details which are missed by the naked eye. This could help brands fight against the plague
of retail copies and also protect the consumer from spending money on something which is below their expectations.
The financial sector will benefit from
computer vision services as much as others. The main applications include:
- fraud prevention based on irrefutable identity verification, like iris scanning
- substantial cost savings in retail due to a lower headcount
- faster and more accurate processing of insurance claims
- innovative ways of identifying investment opportunities
- more comfortable and safer means of accessing bank accounts without the need for a card
- the ability to detect fakes and warn the seller or the buyer
These are just the foreseeable uses so far, but most likely as the technology becomes more widely available new ideas will appear, even from other domains like healthcare or security.