What is computer vision? As a product manager, you’ve probably heard a lot about this technology, but if you haven’t built any CV products, you may not understand what it is.
Computer vision is a field of AI that trains computers to interpret the visual world. This means using digital images in tandem with machine learning models to identify and classify objects. Think of computer vision as an alternative for a pair of human eyes verifying what an object is, and where it is.. (Ex: That’s a cat! And here’s exactly where it is in this image…)
Sending humans on-location to find and count objects is expensive, so computer vision has a ton of practical uses in the real world. If you’ve never worked with CV in product management, it may be difficult to imagine how this technology could be leveraged.
Here are 4 real-world product use cases that use computer vision.
Note: I’m actually going to stay away from autonomous driving use cases, since they are the most prevalent and obvious for computer vision (and also the most discussed!).
1 - Looking for defects on an assembly line
In modern manufacturing, defects need to be called on in real time. (Think thousands of teacups coming off a production line, and needing to identify which ones are cracked.)
As products roll off the production line, computer vision can assess images (or videos ) and flag products that are defective. A human may then take a closer look to verify the object is defective.
This can save manufacturing facilities lots of money in offloading this tedious work to machines instead of humans.
2 - Identification of brain injuries on CT-scans
A lot of health conditions are time-sensitive, and doctors have to make very precise - and quick - decisions that are often life-critical. CV is being used more and more in the field of radiology, or the reading of scans of the human body.
It can be very difficult for a human doctor to have the attention-span and decision-making necessary in time sensitive situations. Especially if they’re tired… or on vacation. A doctor can be assisted by computer vision to analyze brain scans quickly. CV can identify possible strokes and other brain injuries quickly in cases where treatment needs to be immediate.
3 - Surveillance in a casino
Security is super important to casino owners, and with computer vision, casino security personnel are able to accurately identify customers when large amounts of money being exchanged.
It’s extremely difficult for security to monitor hundreds (or sometimes thousands) of simultaneous video feeds at once, looking for sketchy behavior. However, a computer vision model can identify possible suspicious activity in real time, which can then notify security peeps to take a closer look.
This use case highlights how computer vision is most commonly used - typically the goal isn’t to definitively identify a security risk, but rather sift through a giant haystack of data (images) to identify possible needles that need a closer look.
4 - Crop Insurance for agriculture
Most product managers don’t think much about agriculture use cases, but it’s an important industry for many AI/ ML use cases, and there’s a ton of opportunity in the ag space.
Imagine a drone flying over crop fields and collecting HD imagery, with the goal of analyzing soil conditions to estimate crop yields for the season.
The same drone could fly over a field after a bad storm or a heat wave, collecting images used to detect damages to crops. A product manager can take the results and surface in a map to show exactly where damaged areas are, and farmers can address the damage in those areas of their crop fields.