The day, my son built my classifier future

I was sitting in the plane from Stuttgart to Berlin, when I rad mails and newspapers just a minute before starting. My messenger triggered a new message. Julian wrote: "Hi dad, when you will arrive? The classifier is ready..." 

Rolling 4 days back, I explained a bit about our recognition framework at our very new company Ryde and the backbone from my established startup snap2life.de and tried to give a small introduction on image recognition and the power of content recognition. snap2life.de was a result of 10 years research and many products, POI, AR and pattern recognition with openCV, before the time comes by.

And now, my son. A young guy, learned python and raspberry pi and some scripting in go, assembler and now tensorflow. 

He recognized, that we like to extend our clustering classifier for more objects and gestures and took a chance. He searched a few pages and trained some objects. He combined the classifier, rad about random forest, convolutional neural networks and started to mix.

It took the flight and a few moments after to touch my senses and get a real feeling about. After I entered my house, he waited with a smart smile and had this gangster grooming behind. The meaning of power and kingdom and proudness. I came over and had a look into and I am - could say this - very proud and a little shamed as well.

What a time and change, if a clever one could learn all these rules in a day or three. What happened to the complexity of the old days of CUDA processing and optimization, compiles over nights and the research projects for pattern and regression detection with c++ and assembler within. 

Train and discover, train and explore, train and combine and use! What future will be the one, when mathematics, physicians, students and high school people are able to model and cluster a world, spin agent nets and visualize in D3 at a browser in a Sommoment.

Sure, to paralyze and bring it into Kafka topics and Spark Streaming jobs on big data clusters is a different story - but how long?

We as entrepreneurs, big data architects or engineers should not hold so hardly on our crusted experiences, if they are invented by easy machine learning algorithms based on training power and easy regression trees for decisions in microseconds...

Yes, implementation on fastCV variants, CUDA, reflection detection and object tracking for augmented and virtual reality are another approach and yes, Metaio or vuforia made a good job, as we at the DAI and snap2life.de are doing this funny details below the openCV, tensorflow, caffe, theano and deeplearning4j libs. But we should be aware in the new behavior of data understanding and handling.

Many thanks to the open source communities and the strong will to change the world. Please let's be careful in that so fast disruptive transformation process and thank you for this moments in our family partnership, I am so proud of.

Julians dad.

PS: Apples image recognition is telling me, that this article's image is "crema". So, a bit to do, but hey, you have your genius youngsters as well in US, I am sure.