Brodmann17: A Revolutionary Deep Learning Vision Software for Embedded Devices

Adi Pinhas

Machine learning stands out as a transformational technology in today’s data-driven world. Deep learning, a subset of machine learning continues to transform every industry and is significantly helping enterprises to bring a new breed of innovations. In the automotive industry, this revolutionary technology has become extremely popular in image interpretation and autonomous driving. A company which leverages deep learning and computer vision to deliver the most efficient and accurate embedded solutions is Brodmann17.

Brodmann17 has truly expanded its horizons and has grasped opportunities which are not only ground-breaking but also a contributing factor for the deployment of innovative technologies.

Brodmann17 is the leading provider of software-only perception technology for Advanced Driver Assistance Systems (ADAS) and Automated Driving (AD). The company’s solution exceeds current state-of-the-art accuracy while consuming just a fraction of computing power, enabling the transition of these AI capabilities from premium vehicle models to mass-market models. The technology is pure software and hardware-agnostic, allowing Tier-1 suppliers and automakers to combine any visual sensor with any processor across the range of L1-L4 automation.

Brodmann17’s detection software suite utilizes its unique deep learning technology to provide an accurate interpretation of camera video to identify road objects in complex landscapes. It detects moving objects, such as vehicles and pedestrians, static objects like traffic lights and traffic signs, lane detection, etc.

Brodmann17’s solution is packaged into different products for the automotive industry: Aftermarket and passive ADAS, rear camera and surround ADAS, and front camera active safety capabilities.

 

Driving A Quantum Leap in Deep Learning

Brodmann17’s co-founders have been working on deep-learning since early 2012 at various companies when they realized that deep learning was the best algorithm for computer vision applications. They knew that the next frontier was taking this technology to the edge.

With ADAS projected to reach 80% adoption by 2025, automakers and Tier-1 suppliers are now looking for practical solutions to fulfill the demand for deep-learning edge-processing. Brodmann17 is dedicated to providing this technology to the automotive industry to make it accessible to the mass-market and ultimately, save lives. To solve the challenge of being efficient without compromising accuracy, the company essentially reinvented deep learning, resulting in a 95% cost reduction in the calculations needed to produce accurate computer vision capabilities.

 

The Visionary Leadership

Adi Pinhas is the Co-Founder and CEO of Brodmann17 and is recognized as an Inc. 500 CEO. He previously served as a research team lead at Intel and founded two successful companies – surveillance video recording company Vigilant Technology and JustVisual, a large-scale deep learning visual search engine. He holds an M.Sc. in electrical engineering and computer vision.

 

Augmenting AI and Deep Learning

The only drawback of AI and more specifically, deep learning is the amount of processing power it requires. The two main approaches today to address it is either creating a more efficient processor or to try to optimize the algorithms.

In terms of the latter, companies and researchers are modifying existing deep learning open sources, as inventing new deep learning algorithms is very risky and highly complex (usually done by the global corporate research centers or universities). But at Brodmann17, the company realized that modifying existing algorithms can be very limited, and soon enough the neural network starts to lose accuracy. The only way to approach this challenge is by inventing a new deep learning algorithm from scratch — one that is optimized for the edge.

Brodmann17’s novel approach is proven to work and now it is offering the same to automakers as a means to bring better products to market in shorter time frames.

 

Delivering Exceptional Performance

Adi believes automakers are looking to close the gap between what people will pay for these features and the cost of implementing them. They need practical solutions that utilize the low-power processors and other hardware they already have on hand without compromising on accuracy. “With our unique software-only deep learning technology, we can offer 20x improvement on performance from any type of hardware, including low-power processors. This is a key differentiator that makes us stand out, as customers can be resource-efficient while still offering accurate ADAS features to the mass-market,” Adi asserts.

The innovation was developed in-house by the Brodmann17 team of machine learning engineers and researchers.

 

Harnessing the Power of Artificial Intelligence

Adi said when deep learning came on the scene several years ago, people realized its massive potential but struggled to find the best applications for it. Companies in a variety of industries are trying to keep pace with global giants who are already ahead of the curve and implementing it. “In the next few years, we’ll see that AI-focused startups like Brodmann17, who have invested many resources to develop this technology and are providing ground-breaking solutions, will serve as crucial partners for companies looking to tap into the power of AI and level the playing field,” he highlighted.

 

Challenges & Hurdles

While hundreds of companies use AI, Brodmann17 is among the few that are actually developing it. To do that one needs to hire and cultivate a top-notch, talented team. The company is always looking to work with the best minds in the field.

Brodmann17’s next challenge according to Adi is to productize the technology for automakers. Bringing AI to vehicles is new to everyone and poses an interesting challenge.

 

Business Excellence and Awards

Brodmann17 is proud to have reached significant and noteworthy milestones so far. A few of the notable ones include:

•  Brodmann17’s deep-learning technology is proven with different PoCs and benchmarks to provide the same or better accuracy with 5% of the compute power.

•  Brodmann17’s accuracy is equal to the top 10 KITTI benchmark champions and the only one that runs in real-time.

•  The company has raised $11milllion in Series A funding in March 2019.

•  It has won the Red Herring Award in 2019.

 

The Future Ahead

Adi said in the short term, the real revolution is ADAS and AD features, such as automated cruise control, automated emergency break, lane keeping, etc. These are functionalities that promote safety and comfort and are widely desired by drivers. The challenge for the near future is to make them even safer and transition them from premium vehicle models to mass-market models.

“Once that’s achieved, we’ll start to see the more advanced levels of automation been deployed widely,” he mentioned.

No Comments so far

Jump into a conversation

No Comments Yet!

You can be the one to start a conversation.

Your data will be safe!Your e-mail address will not be published. Also other data will not be shared with third person.