Top 5 Computer Vision Use Cases for Consumers
Leveraging the ever-increasing amount of image and video data being collected in this day and age, the field of computer vision is one of the most high-growth research areas within machine learning today. Computer vision entails teaching algorithms to “see”, or make inferences based on the images it is trained on. The applications of this kind of technology are truly countless, especially as part of everyday life for consumers.
When it comes to those applications, most people would think immediately of self-driving cars or robots. But there are many more use cases for computer vision than most people think, spanning a large number of industries.
In 2021, we talked to over 100 computer vision companies. Here are five of the most interesting or innovative B2C use cases we have seen, and some of the companies behind them.
1. Clothes-fitting
Brarista is a startup based in England that is using computer vision to accurately fit bras for women, solving “one of the biggest problems of female well-being”. Their deep learning model for bra-fitting can be used easily at home via a smartphone app. It’s made in direct collaboration with expert eyesight bra-fitters. They collect their data from volunteers.
Regna also uses computer vision to take body measurements. Their solution performs a scan of the body, generating a set of measurements used to accurately fit performance wear.
2. Sperm analysis
Mojo uses dozens of deep learning models trained on over 100 million sperm images to analyse sperm samples from their customers. Their models measure critical sperm parameters to give helpful analyses of a customer’s fertility and overall reproductive health. Their data-powered computer vision solution makes these kinds of insights quicker, more convenient, and more accurate.
3. Sports & fitness analysis
Vay uses computer vision for human motion analysis. Specifically, their solution pursues a computer vision task called human pose estimation. Simply put, identifies and tracks the major joints in the human body through visual data, creating a computer model of the body and how it moves. These models are applied to fitness tasks and allow the user to explore a lot about their exercise- range-of-motion analytics, rep tracking, speed, and mistakes (or sub-optimal executions).
Jogo enables live performance analytics for soccer players using computer vision. They’ve built algorithms to recognize the ball and the player live within their app. The solution can be used to track measures like footwork, agility, and ball control, ultimately analysing player performance and tracking areas for improvement over time.
4. At-home health screening
SkinVision has created an app that uses your phone camera to detect skin cancer. Working directly with skin health professionals, they were able to develop a computer vision model that can accurately detect anomalies, warning signs, and other skin health indicators. It also offers users personalized skin health advice and recommendations based on the visual data it collects from you. The goal: make sure its users visit a medical professional at the first sign of skin cancer, allowing users to be proactive and take their skin health into their own hands.
Zaamigo is a Swiss startup aiming to “democratize dentistry”. They’ve created a toothbrush-like intra-oral camera that you can use to do dental screenings in your own home. The video feed from the camera generates data used to analyze your teeth and gums. Using computer vision algorithms trained on dental data, the app offers you important dental health insights. For example, it can identify calc, stains, and inflamed gums, as well as inform you of dental issues that you should contact a professional about.
5. Car inspection & appraisal
Tractable uses computer vision to assess damage to and inspect cars. Their technology allows users to upload photos and videos of their cars for a variety of purposes, the most significant one being inspection. It can offer valuations of cars and also assess the extent to which cars have been damaged and make estimates on costs. Their algorithms are trained on millions of images in order to make accurate appraisals and reports while upholding standards. While a human appraiser might take years to learn what to look out for, Tractable can train their AI in just a few days.
Images via social media and company websites.
Note: This blog post was not sponsored by any of the above companies. We just think what they’re creating is pretty cool!
At Lightly, we are also innovating in the field of computer vision by helping companies like the ones listed here do what they do even better. By using state of the art methods like self-supervised learning and active learning, LightlyOne curates vision datasets to remove redundancies, cut costs, and improve model performance. If you are curious to learn more, contact us here.