Chooch AI Develops Agnostic Computer Vision Platform

January 19. 2021. 5 mins read

If you had to pick the one technology under the broad umbrella of artificial intelligence that has had the most impact on business and society today, it’s probably computer vision. It’s certainly the most mature, with the earliest research dating back to the days when Elvis first became a hip-swaying phenomenon. The original machine vision applications involved specific use cases for object detection in manufacturing. The term “computer vision” normally refers to more recent synchronicity between developments in big data, the exponential boost in computing power, and machine-learning algorithms. Advances in these and related technologies mean machines can do much more than spot defects in a widget. One AI startup, Chooch, claims its computer vision platform can see and do it all.

Computer Vision Applications – Abbreviated Version

We’ve certainly covered our share of computer vision startups in the past. The technology, of course, can be used for good or evil or marketing in videos. Facial recognition has become scary good, especially if you live in China. Computer vision is one of the key technologies that will put self-driving cars on the road. Algorithms can now ingest imagery from drones and satellites and then provide agricultural intelligence on the next big blight. Medical diagnostics is primed and ready for disruption by AI that can spot tumors or perform ultrasounds using smartphones. You can even invest in such companies, though most are machine vision stocks, but the IPO frenzy in computer vision-based companies is picking up. 

And on and on it goes. And so when we say that Chooch claims that it’s developing an agnostic visual AI platform for most if not all of these solutions, it’s a big order to fill.

Chooch AI: An Agnostic Computer Vision Platform

Click for company website

Founded in 2015, Chooch is nominally headquartered in San Francisco and has raised $25.8 million, most of which came late last year in a $20 million Series A. Crunchbase lists more than 30 investors, with an international flavor, including venture capital firms from New York to Istanbul to Singapore. The breadth of interest probably reflects the diversity of industries the company is addressing: geospatial, healthcare, security, media, industrial, retail, and the U.S. government. It’s also worth noting that the company employs AI hardware from the world’s biggest AI chipmaker, Nvidia (NVDA), which also provides special support to Chooch as part of its various startup and accelerator programs.

The Chooch Visual AI platform.
Credit: Chooch AI

It sounds like Chooch needs all the firepower it can get to deliver a range of results including everything from tracking surgical procedures to facial authentication. The AI platform works on both the cloud for heavy computing and on edge devices, with the company playing up how the system is hardware and software agnostic. Agnostic doesn’t refer to someone in a constant existential crisis. In this case, it refers to how Chooch can process any imagery such as visible light, electro-optical, and x-ray sourced from sensors and platforms ranging from smartphones to drones, medical imaging devices to infrared cameras. The company says its platform, by understanding context, can select the appropriate AI model from more than 2,400 it has already developed for its customers. That provides faster results with higher accuracy.  

There’s Even a Free Demo

Chooch doesn’t expect prospective clients to take all of this at face value. In fact, the company goes to great pains to explain the functionality behind its computer vision services. And there’s a free app where users can see demos of the tech at work. For example, you can stream video from a smartphone to the Chooch server, which identifies objects and adds red-tagged metadata, like in this example: 

Examples of tags generated by Chooch AI computer vision
Chooch AI claims its platform can be used across a wide range of industries. Credit: Chooch AI

Or you can also upload a photo to the web-based AI demo and have a little fun: We chose this mountain scene from Colorado, which the Chooch platform tagged:

Photo tagged by computer vision.
Credit: Nanalyze/Chooch AI

It definitely figured out there were trees in the scene, but its guess as to the location – Clingmans Dome in North Carolina – was way off. It also identifies one of the trees as a Christmas tree, and we really hate Christmas. Also, no mention of the small lake or the clouds and mountains in the distance. But, of course, the more users engage with the platform, the better it will get in theory. In the app, for instance, you can train Chooch to do object recognition. There’s even a live object recognition feature. For a person, the AI generates tags like age and fashion, so for one of our MBAs the results might be “really old” and “absolutely no fashion sense.” Better self-esteem through technology.

Use Cases For Chooch AI

TechCrunch reported that the company has 18 enterprise clients. Here are some example deployments from Chooch AI:

As we all know by now, the coronavirus has infected every market imaginable, so it is not surprising to see Chooch jumping on that train. The startup has launched a suite of AI solutions with its computer vision platform to detect social distancing, coughs, masks, hand washing, and even fevers. While it’s probably not something you’d want to deploy at the next Covid swinger party, the company is selling the idea as a public safety measure. 

People wear facial masks.
Definitely not the Los Angeles Covid swinger party. Credit: Chooch AI

On the diagnostic side for its COVID-related solutions, Chooch AI has been trained to recognize signs of lung injury from x-rays, CT scans, MRIs, and ultrasounds.

One of the company’s enterprise clients is SIPA USA, an international photo agency that uses Chooch AI to tag photos automatically. For example, SIPA trained Chooch for fashion photography, so the computer recognizes dress features and other attributes – something other fashion-forward AI startups have been doing for a while. SIPA also uses Chooch to tag photos from sporting events and big self-masturbation celebrity events like the Oscars where it needs a rich trove of metadata for every star and starlet. It costs 80 cents per image for a human to do the same work that Chooch can do for a penny. 


Here’s our two cents on the Chooch AI visual platform: If the company can deliver a scalable, easy-to-use solution that works at that sort of price point across all those industries, then we would expect to see some rapid growth. In fact, the latest cash infusion is earmarked to hire more data scientists and begin global marketing efforts. Still, it seems like a lot to deliver, especially in areas like healthcare where getting the green light from the FDA for medical use is a key metric for success. We’d want to see some clinical research before we turned over our MRI to Chooch. The company has been around for almost six years now, so we assume most of that time has been spent on R&D. Now we’ll see how well all the work pays off.


Leave a Reply

Your email address will not be published.