Embracing automation with intelligent cameras and how to do it in your case
Computer Vision is nowadays one of the most popular term in the collection of all Artificial Intelligence related buzz and technology words. Everyone has heard about its applications like face recognition, automotive vehicles, bar codes readers, surveillance cameras, visual quality inspection systems and the list goes on and on. Most of us have used the technology already at least a dozen times. And if I was to define it with just one sentence, that would probably be something like:
Computer Vision enables machines to see.
And indeed, Computer Vision transforms all sorts of computers into intelligent machines by enabling them to identify objects, analyze scenes and activities in real-life visual environments. This can happen in real-time (i.e. cameras analyzing production lines, packaging processes or inspecting quality, vehicles detecting obstacles to improve safety etc.) or offline (i.e. large collections of video / images analytics count objects, detect hazardous or illegal activities in drones footage etc.).
It is worth mentioning that Computer Vision is not only about automation. In other words, to do things faster, eliminate human error, and eventually offload humans from boring and repetitive stuff. Beyond that, it very often is applied to detect microscopic level defects that cannot be seen or properly identified with human eyes. And in that sense it augments human capabilities and helps us bring the quality across industries to a completely new level.
But how to start with Computer Vision in your industry?
Our team at byteLAKE together with Lenovo have worked out a model to make it as easy as possible for our clients and partners to successfully implement Computer Vision. And most importantly, to see the results and experience first tangible benefits within weeks rather than months or even years. We are with our clients and partners along the complete journey: from early inception, thru proper selection of proof-of-concepts, all the way to solution delivery. And this is how we do it:
- We can start as early as when the ideas of using Computer Vision are quite vague. Thru a series of workshops, we advise about technology and help understand where it is best to apply it in your particular scenarios. And I must say that we speak human language and organize exciting, funny and energizing workshops! Tempting enough?
- Later on or if you already have your ideas, we can help you transform them into first proof-of-concepts really fast. But before we do it, we will also help you evaluate them and share our opinions as well. We have been working in the field for years, established key partnerships along the way and also built a number of starter kits to jump start any vision related project quickly and deliver results efficiently. You can find some of our starter kits on our website under Computer Vision section.
- We have all heard that Computer Vision is about processing a huge amount of data, deep learning algorithms are complex and require astronomically powerful machines and so on. But… we have got your back once again! But first things first… We know how to optimize such algorithms for powerful machines. We have been doing it for years, helping our clients get the results faster and ultimately cutting down their total cost of ownership. Optimization = making the most of your hardware, reducing time of using most expensive resources, lowering energy bills and possibility to do more with your hardware. But then you need hardware anyway… And here is the point where you need to check out Lenovo’s TruScale, The Pay-for-What-You-Use Data Center Solution.
So what’s next?
Meet us during the upcoming AI Summit in San Francisco, CA (The Palace of Fine Arts, Sept. 25–26) @ Lenovo’s booth. We’ll show Computer Vision in Action, analyzing drones footage. And obviously there’s more than just algorithms analyzing images. Meet us at the booth as we’ll be sharing a complete story of how we helped collect the data properly, analyze it with subject matter experts, prepare first data sets, train the models…
…and later on reduce the training time from weeks to hours, something I mentioned in point 3 above. See you there?
