The world becomes more and more digital, and we are overwhelmed by a digital video content explosion, yet large scale processing of video content remains a huge challenge.

Thanks to a relatively new technique called ‘deep learning’ (read more) we start to automatically derive semantic information from complex sceneries.

We at Robovision are working hard to create the required hardware + software for processing this abundance of visual information. We want to share our expertise with you and provide the tools necessary to enable AI for your business. We transform your data and turn them into a valuable asset.

We currently focus on 3 verticals to better focus on the specific aspects of these fields:

More about us

Why choose Robovision as your partner?

  • Proven track record of profoundly innovating verticals (check our use cases), with a worldwide customer base (USA, Brazil, Japan, Australia,…)
  • We provide the complete pipeline. From data preparation to cloudification of your model.
  • With a solid team of 11 leading experts we can offer a wide variety of deep learning techniques and architectures.
  • We have an extensive background in machine vision and image processing.


Deep learning enables companies and society to analyze and interpret complex videos and pictures by using vast amounts of virtual neurons stacked in layers on top of each other. Read more about what we offer.

Learn more


For specific tracks Robovision works together with Robovision Integrated Solutions to combine the power of AI with the next generation of robots.

Learn more

“Robovision enabled us to create a genuinely interactive experience. Deep learning enables the robot to see  and interact with our employees.”


Patrick Danau, CEO AUDI


We always start with an analysis of the use case. You are the specialist in your business, so together with you we analyze and format your data, to prepare it for the deep learning engine. We create a perspective of what a future data backbone will look like in your business.

Our team investigates how to manage this pipeline in your particular case.

This phase usually takes 2 to 3 weeks.

In the second phase we adapt the RVAI (RoboVision Artificial Intelligence Engine) to the particular needs in your vertical. This means that we train pilot datasets, gradually facilitating the access to your team members.

In this phase we also start to use data augmentation (a type of multiplication of the dataset in order to limit the data labelling time).

In parallel, we discuss a custom, tailor made framework, compatible and competitive with your business model. This is exactly why we only focus on three verticals: we want to maximally align the incentive with you, together we will turn it into a success.

This phase usually takes 1 to 3 months, depending on the complexity of the data.

Now the real fun part starts. The tests on large scale datasets improve on a daily basis and our joint action teams start planning the scale up phase. Thanks to powerful partners as IBM and Amazon AWS and our proprietary parallelization schemes we are ready for prime time. As a privileged partner of IBM PowerAI we have the right contacts worldwide for quick upscaling and large scale integration.

This phase can take 1 to 2 months.

The backbone is now solidly growing, and needs a proper face. Depending on the particulars of your business, we try to compactify the solution into an embedded device, or make custom user interfaces intended for B2C use. We also work together with partners to develop special apps that interface with RVAI.

This phase usually takes 1 to 2 months.

We aim at delivering products with minimum support needs. In this phase we handover to you the key to our turbo engine. We polish the last interfaces and train your support staff.

Not to worry, we are still there to help you out, but the RVAI backbone is that solid you will hardly need us.

This phase usually takes 2 to 4 weeks.

Now that our first success is growing in the market, our teams are more and more connected and your organisation has benefited from a new data-oriented paradigm. We start thinking about a new innovation cycle (step 7).

Building on this success, let’s go to the next endeavor. Our respective teams meanwhile are well integrated and we have both learned from our past mistakes.

This next challenge will go even smoother.




Robovision is at its best when we can serve large robotic farms in an industrial environment. We usually work with one powerful GPU machine (local cloud or ‘fog’). 

read more


Agriculture is a very good match for AI and deep learning.
There are so many types of crops that it is impossible to keep doing 
heuristic programming.

read more


Robovision specializes in human behaviour detection in video streams. From safety applications to the processing of a television stream. 

read more



To assemble a car, the car manufacturer needs a solid and validated flow of components. High tech scanning technology can guarantee that a specific compontent will arrive just in time on the production line. Audi relies on Robovision to fulfill this need.

By applying 3D deep learning on manufactured pieces Robovision checks if the right subtype is selected, and if a set of components is complete for a specific car type. In this case RVAI ensures that the WZB (the kit you use when you have a flat tyre) is complete and validated. Thanks to the strong GPUs from NVIDIA we keep the cycle time fast and snappy.


Pepsico manufactures and packages dorito chips in the highlands above Mexico City. With package branding changing so frequently, a powerful AI based central system is needed to teach branding very fast. The RVAI is the solution here. Reflective, blown up and unpredictable, recognizing orientation and branding at high speed is not for the faint hearted in this application.


The next horizon in our agricultural disruption is 3D. Natural products have a tendency to be highly unpredictable, in contrast to manufactured pieces. By configuring convolutional neural networks to 3D data we want to help the industry automate the handling of difficult products. A Jumbo 747 coming from Africa flies frequently to Amsterdam, full of Chrysanthemum stems. We want to grow them here in Western Europe and manipulate them with smart robots (that pick the stems from the mother plant. By doing so we can use the fertile ground of Ethiopia instead to feed the hungry in the region. The AI revolution can generate powerful secondary effects, disrupting the flow to cheap labour and creating a more harmonious world order.

Deep learning applied to the process industry

We apply deep learning to processes. By mounting a series of proprietary sensors on machinery, we are able to generate data that are used as input for a deep learning architecture. In this way we can teach the system what normal stationary behaviour is and how to properly detect anomalies.

More cases


The future of television, deep learning applied to daily shows.

The future of television, deep learning applied to daily shows.

Our world becomes more and more interconnected and our buying behaviour more impulsive. If we watch a show and we are triggered by a nice.. Read More →
What are the principles behind big data and deep learning?

What are the principles behind big data and deep learning?

Deep learning is all about big data and configuring virtual neurons to adapt to a particular dataset. You could say that a deep architecture is.. Read More →



Scanning behaviour, with Movidius (intel) inside

With select Belgian partners (BNP, BPost), Robovision is rolling out AI based 3D behaviour scanning applications (see below for a product image). They are used to prevent.. Read More →

Elon Musk on AI regulation

Is AI a ‘fundamental risk to society’, as claimed by Elon Musk on the National Governors Association summer conference (15 July 2017)? Elon Musk is.. Read More →

RVAI presented on the Cultivate 2017 fair in Columbus, Ohio

Cultivate Ohio is known for its focus on innovation in agriculture, and the learning capabilities of the Robovision AI engine will be presented there by.. Read More →