Advances in deep learning and neural networks have provided enormous advances in natural language processing and computer vision and have the potential to solve big problems in manufacturing, retail, supply chain, agriculture and an countless other business segments. Naturally, tech startups are behind some of the biggest innovations.
Let’s take a look at “applied artificial intelligence (AI)” startups. These are companies that are applying different techniques – be it the processing of images, text, audio, video, categorical or tabular data or combinations of the above – to face various challenges in the industry, from the fulfillment of the promise of autonomous cars until the limits of agricultural production have been exceeded.
We have already arrived? It seems we’ve been waiting for the promises for years, but the work on autonomous driving technology continues. Argo AI is a company that aims to become the complete platform for self-driving vehicles; it encompasses all the software, hardware, maps, and remote infrastructure that will be needed to get us into the glorious future where we don’t have to be on a bus or train to read a book on the way to work.
In collaboration with partners such as Ford and Volkswagen, Argo AI is pushing the boundaries of the investigation, and has just announced Argo Lidar (light detection and ranging), a new approach to perform object checks up to 400 meters away, in addition to perform well at night and in low light conditions and be able to handle transitions like tunnel exits that can cause problems for other lidar arrays (and let’s face it, us poor humans too).
Argo AI doesn’t make wild promises about its current technology, but it seems like it’s putting in a long, hard job to achieve a safe assisted driving experience. It is currently testing its technology in six cities in the United States and is also scheduled to do tests in Europe later this year.
Ceres Imaging
It may not be as ingenious as autonomous driving, but the technology Ceres Imaging is applying to crops can help lower our grocery bill long before we can jump into an autonomous car and have it drive us to the grocery store.
Ceres Imaging offers a wonderful mix of old-school and cutting-edge technology, bypassing satellite or drone images and replacing them with high-resolution cameras mounted on fixed-wing aircraft, and using those images as input to a series models to provide critical information to farmers, such as discovering irrigation problems two to three weeks before they are visible in the field, correcting over- or under-irrigation situations, and calculating how solving these problems will affect yields.
Additionally, Ceres Imaging can relieve farmers of the burden of simple and time-consuming tasks such as tree counting, making this task possible from aerial imagery. Ceres delivers a report in which the number of trees by variety is counted and the locations of those that are missing or damaged are noted, even generating the nursery order for their replacement. It’s just a small example of how AI techniques are enabling advancements even in areas that don’t immediately come to mind when someone says the words “neural network.”
Landing AI
Founded by Andrew Ng, co-founder of Google Brain and former head of data science at Baidu, Landing AI is an attempt to bring the power of AI to sectors that have yet to see the breakthroughs it can bring. The company’s first product, LandingLens, is an integrated platform that allows manufacturers to combine their expertise with Landing AI to produce a continuously improving visual inspection platform. In addition to the manufacturing world, Landing AI is also working on visual inspection systems for the agriculture and automotive industries.
An interesting aspect of the Landing AI approach is that it puts user data at the center of the solution. Dealing with input data is often the least exciting part of a data scientist’s job, but despite the great strides that have been made in self-monitoring solutions in recent years, it is in the input data that you have more impact the application. It doesn’t matter how sophisticated your model is; If you feed it garbage, you are going to get garbage. This is why AI in emerging countries is focused on easy-to-use and efficient labeling systems, making sure data is collected on an ongoing basis, facilitating retraining and validation of data.
models and, of course, in being able to quickly alert if inferences suddenly deviate (for example, if a camera loses a color channel).
Sentinel
Sooner rather than later, we are going to need a way to detect deepfakes. Although deepfaking – the use of AI techniques to generate fake audio and video of real people – is not yet dominant, the expense and knowledge required to generate these types of actions is decreasing every week. You may have seen the latest news about Tom Cruise’s incredibly compelling deepfake on TikTok. In our future there will be even more compelling fake Tom Cruises.
Based in Estonia, Sentinel strives to be one of the leaders in that field. With impressive NATO cybersecurity credentials and the backing of the former Estonian president, the company offers an API that draws on various deep learning approaches, as well as a massive database of existing counterfeits for comparison purposes, to determine whether the uploaded contents are false or not. The Sentinel system even produces a report on what was done to generate the forgery in the event of a positive result.
Standard
Like the Amazon Go stores that dot some of the major cities in the United States, Standard offers the promise of shopping in brick-and-mortar stores without queuing. The user signs up with a mobile app when they enter the store, stroll through it, pick up what they want, and then just walk away. Standard’s computer vision technology tracks everything that leaves the building and charges it to your account. The experience is even less frictionless than Amazon Go, with no turnstiles or gates.
Standard would very much like to be the company that makes this technology ubiquitous in retail, connecting to their supply chains to provide detailed analytics and enable the most seamless payment experiences. Currently, the company has a flagship store in San Francisco (but sure!) And has signed an agreement with Circle K to conduct some pilot experiments in Arizona, equipping four stores with autonomous checkout technology. If all goes well, we could see Standard’s purchasing AI quickly spread across the country.
