Untitled topic

We have been bootstrapping this thing for over 10 months by now, built on my Bachelor thesis from 2014, revived and Cloud-Nativized. I believe shooting birds with cannons by building Convolutional Neural Network Machine Learning Pipelines for Visual Search purposes can have a #Greener alternative:

Going back to the basics where this field of research started and applying the science of computational statistics by repurposing old-fashioned Image Processing and Machine Learning algorithms to build a classic, Feature-Engineered Computer Vision System. Which scales? Yes.

The challenge was implementing and combining a bunch of algorithms designed and developed at a time when the hardware was limited. The research, optimizations, reiterations of the codes, and the hassle with C++ segfaults in runtime, and in the cloud, was not a simple way to go.

But it has worthed.

Now I can proudly say we have achieved something big: a four-digit-fold competitive advantage in the presumed pricing model compared to the competition, which also means we have a large room for extension and improvements. We also can say we have an MVP System with an accuracy of 96%, with a relatively low response time, below 10 seconds, which can dynamically scale up to even deal with the bombardment of incoming new Objects to Process, Store, Recognize and Retrieve.

We have built a toolset for everyone keen to test and play around with the service: - a Web Console to manage API accesses and access Demo datasets - a Data Explorer app to play with the production API, upload, test, and run recognition on your data - and an API to quickly integrate and build Clients upon.

Links:

Product page: https://www.greeneyes.ai/general...

API Docs: https://console.cloud.greeneyes....

Sign Up to Web Console: https://console.cloud.greeneyes....

My Bachelor Thesis about the theories behind (2014, Hungarian): https://www.arpi.im/thesis.pdf

ANY QUESTIONS, feel free to approach me directly: arpad@greeneyes.ai

Thanks, Arpad