
AI for low-power IoT devices
Together with Lōtik, we created an intelligent low-power wireless sensor that clamps onto pipes and uses on-device signal processing and neural networks to detect water flow and leaks.
Together with Lōtik, we created an intelligent low-power wireless sensor that clamps onto pipes and uses on-device signal processing and neural networks to detect water flow and leaks.
In collaboration with our medical technology clients, we implemented a number of deep learning-based medical image analysis applications for ultrasound and MRI.
For one client, we deployed ultrasound intelligence that was so far only available on high-end devices, to consumer smartphones, while achieving greater accuracy and real-time interaction.
We helped Jetpac, a company that used artificial intelligence to build city guides, implement a deep learning-based computer vision system that finds dogs among other curious things inside millions of Instagram photos. Jetpac used information extracted from images to automatically categorize locations.
Jetpac has since been acquired by Google.
We worked with researchers from Oregon State University on
improving software algorithms for detection and classification of
marine mammals in underwater recordings. Our solution is not only
more accurate than conventional methods used, it's also much faster,
processing years of audio recordings in a matter of
hours.
→ read more
Together with the Museum für Naturkunde in Berlin, we've developed the PhotoID app that can automatically identify hundreds of animal and plant species based on photos.
An interview on ARTE FutureMag demos an early version
(French, German).
→ read more
Natural Vision showed me how to quickly apply their technology to a whole bunch of image-recognition problems we had at Jetpac, and produce much better results than our conventional methods.
Pete Warden, CTO of JetPac Inc
I must say, I am impressed.
Wolfgang Thomas, nature enthusiast, about PhotoID