Vision Systems Design Contributing Editor David Dechow recently assumed a new role as VP, Outreach and Vision Technology, with Landing AI. I caught up with Dechow to talk about the new role, what drew him to it, and his view on where deep learning/AI is going as it applies to vision/imaging.
Mc Loone: What attracted you to this new position?
Dechow: Machine vision has always been a fluid industry with changes in technology and application paradigms that continually drive success and growth in use cases. I’m perpetually interested in new concepts in machine vision, and the merging of traditional machine vision with the use of deep learning tools for inspection in industrial automation is one of the cutting-edge approaches that really has potential to make a difference in applications. The two most attractive parts of my new position are the opportunity to really have some impact on the direction and concept of the Landing AI product in the marketplace and also the chance to work with the truly brilliant Landing AI team and directly with Mr. Andrew Ng who unarguably is one of the world’s most creative, knowledgeable, and influential figures in the area of AI and deep learning.
Mc Loone: What challenges will there be transitioning from the integration side to supplier side of vision/imaging?
Dechow: I’ve always been a “machine vision systems integrator,” though this is not the first time that I’ve worked from the supplier side of the equation. In this case, the incredible upside is that I am coming into Landing AI specifically for my expertise in systems integration and I do not have to “leave that hat at the door.” I will continue to work with the whole machine vision market, including suppliers and value-added systems integrators in my new role.
Related
- Vision Systems Design Contributing Editor David Dechow Starts New Role
- Designing effective traditional and deep learning-based inspection systems for machine vision applications
- Lenses and optics—an integrator’s perspective
Mc Loone: Talk about where deep learning is headed in vision/imaging.
Dechow: In general vision/imaging, the value of deep learning is established. Many applications in varying consumer markets and tasks like autonomous driving really require deep learning and AI. In industrial automation applications, deep learning is evolving and “settling in” as a robust tool that delivers online inspection capabilities in a wide range of use cases. Often, deep learning—when used correctly—can solve automated inspection tasks that simply cannot be addressed with traditional discrete analysis machine vision tools. In other cases, deep learning might supplement those traditional tools to provide better reliability. Even in situations where an application can be solved by either approach, end users might find that a deep learning approach may ultimately be an easier and shorter path to the solution. To be clear though: I believe that we still must apply all the tools properly and in the right applications. Importantly, the need for quality imaging and a well-defined and labeled dataset remain critical to getting the best outcome from deep learning technology.
Mc Loone: Any general comments about the market and Landing AI’s place in it?
Dechow: The machine vision market is, of course, mature and very successful, yet there is room for growth and change. Deep learning technology fulfills a need within the marketplace that has yet to be fully realized. Not to state the obvious, but the number of new companies offering an AI and/or deep learning product for applications in industrial automation has expanded dramatically in the past several years. One must expect that there will be some consolidation or dropout, but a leader ultimately will emerge, and I am confident and convinced that leader is Landing AI. Landing AI has the team, the product approach, and most importantly the incredible depth of knowledge in this technology base to capture a prominent position in the market.
Mc Loone: What keeps you up at night?
Dechow: Usually, it’s about whether or not the grammar and punctuation was correct in the last article or white paper I just submitted! Seriously though, while the mind is always beset by “gremlins in the night,” I try to avoid thinking too much about the things I can’t impact very much while trying to be somehow prepared for them anyway. The machine vision and industrial automation market space has taken many twists and turns over the last four decades, and I don’t exactly know what the next twist will be. Ultimately, I think that Louis Pasteur had it right in observing that “fortune favors the prepared mind.” If we do our best to be prepared for success, then hopefully we can overcome the diverse obstacles that present themselves on the way!
