Computer vision technology (also known as machine vision) allows machines to visually interpret the world around them. As a form of artificial intelligence, computer vision is essentially all about analysing and learning from data – it’s just that the data being processed is visual, rather than, say, numerical or textual. Typically, this visual data is in the form of photos or videos, but it can also include data from thermal and infrared cameras.
With facial recognition being a prime example, the most commonly cited use cases for computer vision tend to be in the fields of security and law enforcement. However, in this article, I want to showcase some of the less obvious uses for computer vision. In particular, the following three industries could benefit hugely from this technology trend.
Computer vision has a range of uses in farming, including detecting weeds, diseases and pests, analysing the land, spotting water leaks, tracking animals, and sorting and categorising produce when picked. All of this can help to reduce costs for farmers, while maximising efficiency and increasing yields.
In one example, computer vision and machine learning are being used to detect the ripeness of papayas. A team of researchers from the University of Campinas and Londrina State University in Brazil has been developing computer vision software that can detect the ripeness of fruit based on images – with an accuracy rate of 94.7 percent. The idaea is to help Brazilian papaya growers maximise the value of their fruit, by selecting less ripe fruit for export and reserving the ripest fruit for sale locally. The researchers also hope to develop a consumer app that would help shoppers buy the right fruit according to how soon they want to eat it.
Elsewhere, Blue River Technology’s See & Spray system uses computer vision to identify which plants are crops and which are weeds, so that individual weeds can be sprayed with herbicides while healthy crops are left untouched. The system reportedly leads to a reduction in herbicide use of 90 percent. Agricultural giant John Deere was so impressed with the system, it ended up acquiring Blue River Technology.
If you think about it, the healthcare industry is particularly rich in visual data, with CT scans, X-rays, and so on. Computer vision allows machines to analyse this visual data and identify abnormalities or disease. This can significantly reduce the amount of time spent on analysing images, thereby relieving some of the strain on doctors and enabling them to spend more time with their patients.
A range of AI-based computer vision tools are being developed specifically for the healthcare sector. One example comes from tech startup MaxQ AI, which has developed software that detects brain bleeds in CT scan images. The detection software, named Accipio Ix, has been approved for use by the FDA, and MaxQ AI has also announced partnerships with Samsung, IBM Watson, and GE Healthcare.
Microsoft is getting in on the act, too, with its InnerEye software, which is designed to identify possible tumours and other abnormalities in X-ray images. Radiologists can upload patient scans; then, the software identifies areas where it believes there are tumours. The radiologist can then focus their attention on scans where problems have been flagged up, rather than healthy scans.
Even setting aside the obvious security applications, there are many potential uses for computer vision in retail. Amazon, for example, has made heavy use of the technology in its small chain of Amazon Go grocery and convenience stores. Thanks to computer vision, Amazon has been able to eliminate the physical checkout process altogether. Once the customer has scanned themselves in at the store entrance (using the Amazon app), they can simply wander around, pick up the items they want, and then leave – without having to queue and pay. Cameras track what the customer selects, and then the cost of items is automatically charged to the customer’s Amazon account.
Computer vision (specifically facial recognition) technology can also be used to identify individual customers in order to give them personalised recommendations and rewards. Upmarket candy retailer Lolli & Pops has been experimenting with such a facial recognition-driven customer loyalty scheme. Customers who opt-in are recognised when they enter the store, meaning sales associates can then give personalised recommendations based on what the system knows about the customer’s preferences (and any allergies).
As computer vision technology is becoming increasingly cheaper and easier to deploy, it’s no wonder the entire market for computer vision is predicted to reach $14 billion by 2024 (up from $9.9 billion in 2019). We can, therefore, expect to see greater use of computer vision across an even wider selection of industries in the very near future.