Deep Learning solutions
for Computer Vision
Our team creates and deploys optimized Deep Learning solutions.
Our smart algorithms recognize people — and can enhance retail CRMs or smart access control systems.
Our Computer Vision algorithms can detect, identify, track and analyze body poses of athletes in wide range of sports — like swimming, soccer or tennis.
Our pose detection algorithms are capable of human body pose analysis— which can be used for applications from automatic workout analysis to AR clothes fitting.
We apply our Deep Learning expertise in medical applications — for example, to diagnose health problems on 3D body scans.
Our people counting platform is trained on large in-house people dataset. It provides state-of-the-art accuracy at 50x smaller model size and 370x less computations. The platform is flexible and can be used at the edge device like NVIDIA® Jetson™ or at a cloud server.
Our labelling team annotates custom dataa
Our in-house tools create optimized neural networks. We engineerd these to deliver the models up to 5 times faster.
We use state-of-the-art pruning and quantization techniques to make models fast even on edge devices. 2x less MACs and same accuracy level.
FPGA MOBILE CPU CLOUD
50 Mb — Average detection model
We will discuss how deep learning solutions can make your product smarter and make time and cost email@example.com