Need autonomous driving training data? ›

9 Computer Vision Predictions for 2019

9 Computer Vision Predictions for 2019

With impressive advances in autonomous vehicles, robotics, drones, retail applications, and much more, computer vision (CV) is in the mainstream lexicon in a way it wasn’t even as recently as two years ago. For those of us who have worked in the field for some time, it’s exciting to see the broader public take notice of the technology that is transforming everything from the way we shop to the way we move about the world.

 

Looking ahead to 2019, here are nine trends in CV that I predict will dominate the year to come.

 

  1. CV applications on the edge. We’ll see CV applications on devices such as phones, security cameras, virtual mirrors, and more become more prevalent in 2019. These will be meaningful for data privacy—an issue consumers are increasingly sensitive to—because the applications won’t have to transfer data over the internet to function.
  2. Customer acceptance of CV to aid health and fitness. Fit Sensors with 3D body scanners, such as Naked by Naked Labs Inc., will permeate more homes as consumers look to new technologies for insights on their health and fitness. These types of sensors can provide feedback such as body fat percentages, circumferences, and track images of the body over time.
  3. CV comes home. Smart speakers that rely on voice-based systems to function have already begun training consumers—including as many as nearly a quarter of U.S. households—to expect when a device is listening or not. Just as consumers have embraced these devices to listen to news, conduct voice-searches, and verbally control systems in their homes, I believe consumers will increasingly accept CV-powered applications in their home—especially when those applications give people control to what it can see and when. Amazon’s Echo Look is one such example, as is Netatmo Welcome, which uses facial recognition technology in outdoor security systems to distinguish between familiar and unfamiliar people approaching your home.
  4. More industrial, supply chain, and inventory CV applications. As manufacturers look for new ways to reduce labor costs and improve safety, they’ll increasingly invest in CV applications that can outperform human vision in assisting in critical manufacturing functions, including predictive maintenance and defect prevention, as well as do jobs that can pose safety risks for humans such as package inspection. We’ll see companies with major logistics operations invest in CV technologies for faster and better distribution. And, we’ll see more retailers invest in inventory robots with cameras, just as Target and Walmart already have.
  5. Broader deployment of cashierless technology. Amazon Go may have gotten there first, but in 2019 we’ll see more competition in this space as consumers increasingly learn to appreciate friction-less checkout experiences. And, Amazon obviously isn’t slowing down; according to Bloomberg, the company may open as many as 3,000 of its cashierless stores by 2021.
  6. Push for data regulations and privacy protections. Thanks to a number of high-profile data privacy scandals, the biggest technology story of 2018 was arguably increasing consumer discontent and mistrust in technology companies to manage their data. Without a doubt, in 2019, we’ll see more calls for congressional leaders to take action in the U.S., just as the EU did earlier this year in implementing GDPR. This will impact CV data in a major way and may lead to stronger requests for transparency and explainability in AI applications (which, IMHO, is a good thing).
  7. CV increasingly aids in developing training data. This is our core business, and I know we’re not alone in investing in CV technologies to make the process of generating high quality training data more efficient and accurate. CV models that provide automated annotations, aid in quality assurance, and better complement human intelligence in developing top notch training datasets will get better in 2019. We’ll also see CV increasingly help in protecting consumer privacy, such as models that anonymize individuals whose faces may appear in data gathered by autonomous vehicles.
  8. More people will ride in autonomous vehicles, and we’ll pass laws impacting them. To-date, first-hand experience of people who have ridden in autonomous vehicles has been limited. While it won’t jump drastically in 2019, expect to see that number gradually increase thanks to services like Waymo One, which launched earlier this month. Additionally, although congress in the U.S. failed to pass autonomous vehicle legislation in 2018, I anticipate it will happen in 2019 as we see greater scrutiny on safety regulations following increased on-road testing in the year ahead.
  9. Climate change concerns will drive more investments in predictive agriculture. CV is revolutionizing many farming tasks, such as planting seeds, harvesting crops, distributing pesticides, water, and fertilizer, and much more. Using CV technologies in predictive farming can help growers farm more sustainably by leveraging technology such as drones to more efficiently manage water supply, sensors that identify growing patterns to maximize yield, mapping tree canopies to protect forests, and more. CV technologies can be a big aid to humans as we continue to grapple with a changing climate and invest in minimizing environmental damage.