AI-powered huupe dreams come true
How Avnet helped usher in a new era of basketball technology
Meet huupe, a start-up company with a slam dunk at the intersection of artificial intelligence, computer vision and the good old-fashioned neighborhood basketball court.
With its innovative products, the huupe PRO and the mini-huupe, the company looks to reinvent how basketball enthusiasts play across geographies and skill levels. By employing machine-learning models trained to various basketball movements, huupe could potentially revolutionize the way professional, amateur and backyard basketball players train and perform.
Think of it as the Peloton of basketball.
Avnet and huupe: A strong partnership
- Provided expert guidance on the system architecture and component selection
- Identified incompatibility and EOL issues with several components, helping to minimize risk and future-proof the design
- Established a global supply chain and expedited component quotes for huupe’s contract manufacturer
- Currently sourcing and buying components for both products
From Snapchat to startup
Smartphone screen shot shows Paul Anton's huupe PRO shot success.
The story of huupe begins with lifelong pals Paul Anton and Lyth Saeed. They grew up in Milwaukee, then Anton moved away to college in Boston. Despite the distance, they still played basketball together via Snapchat.
This simple act of long-distance competition sparked an idea that would eventually lead to the creation of huupe. After developing an augmented reality app for playing basketball remotely via eyeglasses, Anton and Saeed had their a-ha moment during a reunion in Milwaukee in 2019.
Lyth asked, "What if we took the idea and put it on screen on a backboard?"
Anton’s reaction was immediate. "My jaw dropped. THAT was the idea to pursue."
The tech behind huupe
huupe’s flagship products, huupe PRO and the mini-huupe, represent the culmination of years of development, testing and refinement. The huupe PRO, which took five years to bring to market, is a full-sized, weather-resistant basketball hoop equipped with advanced AI and computer vision capabilities. Its smaller sibling, the mini-huupe, packs similar tech into a more compact form factor, perfect for indoor use or smaller spaces.
At the heart of both systems is a commitment to simplicity and functionality. "Not a single thing is out of place or unnecessary. We used the absolute bare minimum tech to make this product do what we needed it to,” Anton said.
This minimalist approach belies the sophisticated technology powering huupe products. The huupe PRO uses a single camera that captures the entire court, paired with a proprietary AI algorithm trained on over 4 million shots in various conditions. This enables the system to track shots with astonishing accuracy, providing real-time feedback and analytics to players.
The mini-huupe, on the other hand, eschews cameras in favor of advanced radar and lidar sensors.
Both systems boast over 99% accuracy in tracking baskets, shot attempts and shot distances, testament to the rigorous development and testing process behind huupe’s products. The data is then relayed to a mobile phone app, where a user can understand shooting trends and use the information to improve skills.
At the core of huupe’s technology is a sophisticated machine-learning model that powers its shot-tracking capabilities. "We collected over 4 million shots in all different conditions, locations, different people. It took us four years to collect that data to create the algorithm," Anton said.
This massive dataset enables huupe’s AI to accurately interpret the complex movements involved in basketball shots, accounting for variables such as lighting conditions, player height and shooting style. The result is a system that can provide instant, accurate feedback on every shot, helping players improve their game in ways that were previously impossible.
The development of huupe required an unprecedented level of technical depth, particularly when it came to component integration.
"We needed to understand on a silicon level how some chips are interacting with other chips," Anton explains. “Avnet facilitated crucial conversations with suppliers, asking probing questions that the huupe team hadn't considered.”
One example of this deep technical collaboration involved the radar digital signal processor chip used in the huupe system. By examining the chip's functionality at the silicon level and its interactions with other components, the team was able to unlock additional computing power, enhancing the overall performance of the product.
The path to production
Bringing huupe from concept to reality took persistence. The team faced several technical challenges, particularly in areas of weatherproofing for the huupe PRO and achieving camera-less shot tracking for the mini-huupe.
To ensure the huupe PRO could withstand the elements, the team pursued and achieved an ingress protection (IP) 55 certification, making the system resistant to dust and water jets from any direction. This level of environmental protection was crucial for a product designed to be used outdoors in various weather conditions.
For the mini-huupe, the challenge lay in developing a shot-tracking system that didn't rely on cameras. The team's innovative approach, drawing on their computer vision backgrounds, resulted in a sensor-based system that matches the accuracy of its camera-equipped counterpart.
Along the way, the huupe team relied heavily on Avnet’s field applications engineers, who advised against several components the huupe designers selected but had no idea were in end-of-life status. Partnering with Avnet enabled huupe to have a clear understanding of sourcing options for each component.
“One of the first things we did for them was a supply chain risk assessment,” said Ali Rahbar, senior account manager for Avnet’s Americas region. “We also attended meetings with suppliers and huupe so that the huupe team had a full understanding of the components’ capabilities now and in the future.”
Bringing a complex, tech-laden product like huupe to market required more than just innovative design—it demanded expertise in supply chain management and manufacturing. This is where Avnet, a key distribution partner, excelled.
"Avnet has been a superstar for us to help us get this product off the ground," Anton said. "The partnership with Avnet proved crucial in navigating the complexities of component selection, availability and optimization.”
One of the more significant contributions from Avnet was in refining huupe’s system architecture.
"Avnet offered guidance on what they've seen fail in the past, how to improve our design and which components to select," Anton said. “This insight was particularly valuable in selecting processing components and designing the overall computer system architecture.”
Avnet’s expertise also proved invaluable in addressing potential supply chain issues.
"Initially, we designed the product with components that were already end-of-life, and Avnet helped us find new components with good supply chain," Anton said. This foresight helped prevent potential production delays and ensured a smoother transition from prototype to full-scale manufacturing.
Beyond basketball
While huupe’s current focus is on basketball, the team is already looking at applications for their technology in other sports. "We've been working on the machine learning of a soccer product. We're also looking at lacrosse and hockey," Anton said.
This expansion into other sports demonstrates the versatility of huupe’s AI-driven approach to sports analytics. By adapting their machine learning models to different types of ball movements and playing environments, huupe has the potential to revolutionize training and performance analysis across a wide range of athletic disciplines.
With the successful launch of huupe PRO and mini-huupe, the team is looking to the future. While they're keeping some cards close to their chest, Anton hints at potential new features, such as voice control, which was initially left out for privacy reasons. Now, it’s being reconsidered due to user demand.
As huupe continues to push the boundaries of what's possible in sports technology, one thing is clear: the intersection of AI, computer vision and athletics is fertile ground for innovation.