The StartUp Beat Daily Email
Enter your email address:


ABOUT STARTUP BEAT
©2007-2012 StartUp Beat/BRK Media, LLC
Powered by Movable Type

« October 2009 | Main | December 2009 »

November 25, 2009

Q&A with Axel Kloth and Royce Johnson, Parimics

As a follow-up to the company pitch that ran earlier this week, StartUp Beat conducted a Q&A with Parimics CTO Axel Kloth and CEO Royce Johnson.  Comments are welcome below!

Parimics

SUB: Who do you consider to be your primary competition?  Is it companies like iRobot?

Royce Johnson: iRobot would be a perfect customer of ours.  They make military and security robots that could use the power of our video analytics processor.  Whenever you have a complex visual environment to deal with, our solution is unmatched in performance.  There are quite a few companies out there providing vision systems and high performance processors.  Some are focusing on niches that do not overlap with our solutions.  Others are focused on general purpose processing solutions that require a much more complicated software solution.  DALSA, Matrox, National Instruments, TI, Intel and Freescale as well as MobilEye are respectable companies with good products, but we believe we have a unique solution to address the challenges of a wide range of vision system problems.  We are the only advanced chipset specifically designed to solve advanced vision system challenges.

SUB: What are your target markets, specifically?

Axel Kloth: Let me expand on your question a bit since it might not be immediately clear where we fit in.  Parimics helps users analyze still images and videos, without having to rely on people.  That sounds abstract, but it is not.  In effect, we enable advanced machine vision.  Most security surveillance systems rely on people actually watching monitors and, unfortunately, break-ins often occur when those guards are not watching.  That method of security is not reliable or efficient.  The same is true for sorting trash and recyclables; that is done by people today.  I would not consider that a sanitary or healthy environment.  I would not want to perform that job.  Jobs like this can be done by machines.

There are many more applications for our technology, such as advanced medical imaging, air traffic control, and automotive driver alert systems.  Considering that every year customers buy about 50 million vehicles, there is a huge market for automotive vision systems that can handle the complexity of typical driving environments.  Cars have become a lot more intelligent over the past few years, and they already avoid lots of accidents with smart driver alert systems.  We would substantially improve upon that.  An advanced driver alert system based on our technology would be able to recognize an impending collision with another car or truck or bus, and alert you to take counter measures before it is too late.  Eventually, our advanced processors can handle the complexity of automatic driver guidance systems.  Imagine the impact on driving safety!

We can also help accelerate drug discovery.  Current technologies cannot rule out toxic drugs fast enough for scientists to move the drugs with good prospects into the next phase.  We can help with that screening process.  Parimics’ technology can reduce the time it takes to run drug discovery image analysis from several weeks to less than half an hour.  Medical diagnostics is another area where we can help tremendously.  An MRI machine captures cross sections of your body.  The data it generates is too much to handle in real-time for current solutions.  We process this amount of data in real-time and allow doctors to see what is hidden inside these cross-sections, not slice by slice, but in a projected 3D-view of your body.

We are currently focused on defense and security applications, and medical image analysis.  But the number of markets needing advanced vision capability is exploding.  We can address problems in robotics, video surveillance, defense, medical analysis, driver alert and control systems, manufacturing quality control, video and image search, and much, much more.

SUB: Explain a little further how your product is a paradigm shifter, as you claimed in your pitch.

Axel Kloth: Parimics’ architecture is novel in its hardware and software approach to image analysis.  We are very specifically targeted towards image analysis and image analysis only.  We don’t run Word or Outlook or Firefox.  Our chipset is not a general-purpose processor chipset.  Special-purpose processors have been successfully deployed in cell phones, in video cards for graphics output, and in other applications such as automotive engine control units.  We have taken this successful approach and applied it to image analysis.  So it is instruction- and energy-efficient, and it’s easy to use and program.  The Parimics platform combines state-of-the-art processors and open-source based software development kits and integrated development environments.  All of these components combined with sample code and two fully integrated vertical solutions will make it very simple to create machine vision applications.

Our processors and the development tools are being designed with ease of use in mind.  In fact, we put effort into making sure that every possible machine vision application can be written or generated with the least possible effort on the side of our customer.  We anticipate that once we have all of our APIs and subroutines done, it will take us less than 12 man-months to implement our first complete vertical.  With four programmers it could take as little as three months to implement a whole new vertical if a customer decides to do so.

SUB: How do you generate revenue?  Are you profitable at this point? 

Royce Johnson: At this time, we are building our initial prototype.  We expect our markets to be very profitable since we are providing our markets with an opportunity to greatly improve productivity and reliability at lower costs.

SUB: How are you marketing your products?  Do you plan to expand your marketing, moving forward?

Royce Johnson: Most of our markets are dominated by a small set of suppliers.  We are contacting market leaders in each market we pursue.  Of course, we expect to expand the number of markets we address as we grow.  At this point in time, we are very focused on getting customer wins in high impact market segments.

SUB: Are you looking for funding?  Do you anticipate looking for funding in the near future?

Royce Johnson: We have been self-funded to this point.  We are now looking for help to complete a very robust prototype and SDK.  Design wins will then make subsequent funding much easier.

Parimics: www.parimics.com

November 19, 2009

Featured Pitch: Parimics

Parimics logo 

Web Site: www.parimics.com
Headquarters:
Saratoga, Calif.
Year Founded:
2002
Founder: Axel Kloth, CTO; Royce Johnson, CEO
Investors: Self-funded
Employees: 5

Company Description in 140 characters or less: 

Parimics builds advanced vision technology that enables any system to make real-time
decisions based on streaming video or image data.

By Royce Johnson 

The world is continually capturing trillions of images that could be harnessed to make real-time decisions.  Royce JohnsonParimics believes that building systems that see and think will take the capability of vision systems to a whole new level.  Video and image data will control systems that run the world’s factories, protect the world’s critical infrastructure, and make all of us healthier, safer and more productive.

The TAMs for systems that require advanced vision capability are currently more than $10B in combined size and expected to grow at least 15 percent per year over the next five years.  Parimics is committed to delivering advanced vision technology that will serve an expanding list of market segments including the following:

• Medical Imaging Systems
• Robotics Systems
• Military and Defense Systems
• Nanometric Manufacturing and Quality Control Systems
• Security and Surveillance Systems
• Automotive Alert Systems
• Video Search Systems
• CleanTech and GreenTech Applications

Customers who need these advanced vision systems do so for many reasons.  Visual data is usually coming too fast to analyze in real time.  The required analysis is very time-consuming since most of the time an overwhelming number of images must be analyzed.  And critical variables can’t be easily seen by humans, making human analysis too costly, inefficient, or impossible. Today’s image analysis solutions are much too slow to provide adequate support for solving these complex customer issues.  Speeds of 20 frames/sec. (FPS) or less are the rule, and probabilistic methods reduce accuracy of current systems. Parimics’ architecture scales from 100 to 20,000 FPS, and our results are deterministic and extremely reliable.  We provide a “super-computer-like” architecture specifically designed for advanced image analysis.  Our scalable approach to solving customer problems easily meets all their performance and cost requirements.  Our goal is to make security and surveillance systems, cars, robots, medical equipment and unmanned vehicles smarter, safer, faster and easier to control and to build.Our Parimics chipset accelerates the analysis of raw video and image files, and then hands that analysis off to higher-level decision software.  Today’s vision systems are greatly underpowered when faced with complex environments.

Our chipset will typically increase the speed of analysis by 100-1000X.  That factor increases as the complexity of the analysis increases. Parimics’ chipset supports a broad set of vision system customers.  Our goal is to be recognized as one of their strategic and premier suppliers. 

Parimics’ Unique Technology

Parimics is building a semiconductor chipset that performs like a super-computer and is capable of delivering real-time video and image analysis.  Our technology is a change in paradigm that overcomes the basic and inherent limits of current image processing technology.  We have been granted two fundamental patents that allow for parallel processing on a pixel-by-pixel basis.  Using an innovative architecture, this technology can perform real-time computation of an object’s velocity, acceleration, trajectory and more.  Parimics’ architecture allows the processing engine to identify, track and analyze an unlimited number of objects in its field of view.  The instantaneous results give the system the power to halt, adjust, identify, notify, analyze, and perform many other functions that will improve the productivity and cost effectiveness of the target system.  We are unique in the following ways:

• We use one processor for each pixel.
• Simple, powerful interconnect architecture is used between all parallel
processors in the array.
• Our architecture combines and leverages image independent and image
dependent analyses.
• Video and image analysis can scale from quarter-VGA to more than 20 megapixel
frames.
• Speed of video and image analysis scales from 100 FPS to 20,000 FPS.
• Software solutions are much simpler and faster to develop on Parimics’
architecture.

Our Founders and AdvisorsAxel Kloth

Axel Kloth is the CTO and architect of our chipset.  He has an expansive physics, systems and chip architecture background.  Axel started his career at Siemens/Infineon where he was instrumental in getting TriCore 1.2 H/W and S/W working for engine control units in VW and Audi vehicles.  He then led the turnaround of HotRail with its switch fabric architecture resulting in a successful acquisition.  Axel architected and supervised the design of iScale, a switch fabric solution that is today used by NTT to handle every phone and cell phone call in Japan.  He also envisioned the first network processor and was the first to attempt to sell the idea on Sand Hill Road.

I am the CEO of Parimics and I have more than thirty years of high tech management experience.  I started my career at Intel and eventually co-led the Intel Microprocessor Division for five years during its critical growth phase from the 286 to development of the Pentium chip.  I managed Intel’s availability improvement task force during lengthy periods of short microprocessor supply, and teamed with Intel’s VP of Sales in executing strategies to keep Intel’s customer base happy during times of tight supply.  I also managed the Intel Vendor of Choice program to its highest ever level of customer service.  I left Intel in 1995 and have been a serial entrepreneur and a key founder and contributor to many startups.  I was a VP of Marketing at Millennia Software, the first company to voice-enable email and faxing capability.  I was also VP of Marketing at ZapMe! which delivered education content via satellite to high schools across America (IPO in Q4 ‘99).  I founded Parimics with Axel in 2002 and have been dedicated to delivering cost-effective advanced computer vision to support a whole new set of market possibilities.Parimics recently recruited three well-known processor technologists to join its Technical Advisory Board.  The new TAB members are John L. Gustafson, Stanley Mazor, and John Wharton.

John L. Gustafson is widely recognized for his research in the speedup of parallel processing systems.  His pioneering work on a 1024-processor nCUBE at Sandia National Laboratories created a watershed in parallel computing, for which he received the inaugural Gordon Bell Award.  Gustafson continues to be at the forefront of high performance computing R&D, and is currently Director of Intel Labs in Santa Clara, CA. Stanley Mazor is one of the inventors of the first microprocessor, the Intel 4004.  He wrote the software for this revolutionary chip.  In 1996, along with his co-inventors (Ted Hoff, Masatoshi Shima, and Federico Faggin) Mazor was inducted into the National Inventors’ Hall of Fame.  After leaving Intel, he joined EDA and CAD startups to help popularize Hardware Description Languages for chip design.  More recently he has been teaching in the industry and guest lecturing in universities.

John Wharton defined the architecture of the Intel 8051 microcontroller, the highest-volume processor family in Intel history.  Since 1980 more than fifty companies have developed software-compatible variations of that device.  From 1989 through 2003, Wharton lectured in Electrical Engineering and Computer Science at Stanford.  He has published scores of articles on microprocessor hardware, software, and operating system design and holds three patents in memory system architecture. In addition, Parimics has established a technical advisory relationship with widely respected Prof. Gerard Medioni, former Chairman of the Computer Science Department at USC.  His department is internationally known for its work on image recognition software, and regularly receives grants and funding to support next generation work in video and image analysis.

Parimics – www.parimics.com

 

November 13, 2009

Q&A with Matt Aird, Co-Founder, Bizwiki

As a follow-up to the company pitch that ran earlier this week, StartUp Beat conducted a Q&A with Bizwiki Co-Founder, Matt Aird.  Comments are welcome below!

Bizwiki logo   

SUB: The site is free for businesses to participate, so what is your business model?  How does (or will) the site make money?

Aird: As you mentioned, Bizwiki is completely free for businesses to list their information on, and also free for the public to use as a search resource.  In the future we intend to add some relevant text advertisements to some sections of the site, allowing us to dedicate editorial resources to ensure the quality and reliability of the site’s information is maintained.  

SUB: Do you pull any Yellow Pages information to populate Bizwiki’s listings, or is it dependent upon businesses proactively adding listings?

Aird: We have licensed some basic contact information about a wide range of companies, giving us a base on which interested parties can build up details about companies.  Any business with an address in the U.S. is invited to add themselves to the site or add more information if we already have an entry for them.

We’ve also built our own custom search spider to help retrieve and update further information about companies, so we are being very proactive about tackling the issue of listing companies with a greater depth from as many angles as possible. 

SUB: Who do you see as your primary competition?

Aird: What we are trying to do is become a primary resource for people who are looking for more comprehensive information about businesses, their products and their services.  Although we are enabling people to publish detailed information for free rather than charging, that puts us in competition with the traditional Yellow Page publishers like Real Yellow Pages, Yellow Book and Superpages, particularly as they feel the squeeze in their publishing businesses and attempt to duplicate their printed books online.  

SUB: It looks like you are getting a good number of businesses listing themselves on the site, so how has your traffic been so far?

Aird: The site’s traffic has ramped up since the beta launch, with well over a quarter of a million people already using Bizwiki.com per month.  We see a great deal of room for growth when the full version of the site goes live next month.  The beta is almost over and the new features and functionality are currently being tested before release. 

SUB: How are you promoting Bizwiki? 

Aird: All our promotion is done online or by word of mouth.  We’re also working hard to make sure search engines can index our site, giving companies that don’t even have a website a chance to have their details found on Google, Yahoo and the other search engines.

Bizwiki: www.bizwiki.com

November 09, 2009

Featured Pitch: Bizwiki

Bizwiki logo 

Web Site: www.bizwiki.com

Headquarters: Aldershot, United Kingdom

Year Founded: 2006

Investors: Self-funded

Employees: 12

Company Description in 140 characters or less:

“Bizwiki.com is the user-edited Business Wiki.  Anyone from Bizwiki’s fast-growing audience can add and edit business listings for free.”

By Matt Aird, Co-Founder

Matt Aird, BizwikiBizwiki is literally a wiki for business.  The site launched in the U.S. in July 2009, promising to change the way local search works by enabling its users to build up the most detailed and up-to-date index of business using wiki-style functionality.

Three years ago myself and co-founder Keith Hinde were discussing our frustrations with the long delay and lag between when a company’s details change and when they have been updated during our work for conventional directory and “Yellow Pages” publishers.  Tens, even hundreds of thousands of important details such as a company’s addresses and phone numbers may be incorrect or out of date, but the books keep on being printed.

In a break with traditional Yellow Pages websites, Bizwiki invites business owners and representatives to get involved in adding and improving their records with everything from contact details to prices and opening hours, completely free of charge.  These can be updated with changes and additional information quickly and easily on the site.

We’re happy to say that the public response has been excellent.  The early alpha-version of Bizwiki.com was opened to the public in December 2008 with the beta following in June, and several hundred thousand people are already using it each month!  There is definitely a strong demand for the sort of information a web 2.0 business site can deliver, and the increasing amount of users on the site provides a compelling motivation for businesses to get involved in adding and editing their listings.

So why should any of your readers on StartUp Beat care?

Bizwiki.com is becoming a detailed and trusted resource, and we are gaining the hundreds of thousands needed to make it worthwhile for people to add and edit the information about businesses on the site.  We plan to build it to the point where small businesses—especially new startups and companies working online—will be driven to add and update their information in order to gain access to free traffic from the site.

That’s not to underestimate the challenges ahead, or downplay the fact that what we have built is the shell the site needs to grow into, but if all goes according to plan you should expect to hear more about Bizwiki.com in the future.

The Bizwiki difference:

• It’s free – Unlike many established publishers that charge for inclusion, Bizwiki is free to search, free to edit and free for companies to list on.
• It’s editable – The “anyone-can-edit” approach is a challenge to the frequently out of date records of conventional printed business directories.
• It’s a wiki – The wiki approach allows far more depth of information about each business to be compiled than anything conventionally available.
• It’s structured – Bizwiki is built using structured data, allowing reusability of information, bulk updates from chambers of commerce or web spiders, and an easy search experience for users.

Bizwiki was built by industry-veterans with years of business directory and meta-search experience behind them, including myself, Keith Hinde, Craig Sefton and Arthur Jenkins, who between them have helped develop directory and search products for Infospace, local directory publisher Thomson Directories, TradePage and Webcrawler.

To try the new Bizwiki site, or even add and edit a business record, visit http://www.bizwiki.com/.

For Bizwiki for the United Kingdom, visit http://www.bizwiki.co.uk/


Hosting by Yahoo!