Computer Vision
The optimal sensing and representation of
our surrounding world are two of the main challenges 7 Sensing Software intents
to solve.
Our experts in computer vision leverage
data from 2D and 3D cameras to extract precise representations of the physical
world.
7 Sensing Software unleashes the potential
of imaging sensors, from the capture to the application, to deeply change the
real-time embedded experience.
Because, in the end, we want to make
interactions between real world and sensed world seamless.
Depth Capture
Our 3D Depth sensing solution is based on Deep Learning models.
It fuses in real-time an RGB image with ultra low-resolution 3D ToF data to create a high-resolution depth map.
Resulting quality is on par with Apple LiDAR-based TrueDepth and outperforms iToF-based and Google’s RGB-based depth maps estimations.
Our technology is designed and optimized for Neural Processing Units : 40ms per frame – 25 fps on Hexagon Processor.
Gesture Recognition / Hand Tracking
Our Gesture Recognition and Hand Tracking technologies are based on Machine and Deep Learning models.
They perform real-time fusion of ultra-low resolution (4×4 or 8X8) 1dToF sensor data to achieve complex touchless UI functions.
Our technology is designed for Sensor Hub platforms (Cortex-M33 @ 150MHz) : ultra-low power consumption on microcontroller to enable always-on operation
Gesture Recognition | Hand Tracking |
66 kBytes Flash 61 kBytes RAM <1 ms per frame | 141 kBytes Flash 127 kBytes RAM 47 ms per frame |
Hand Tracking demo – real time processing on microcontroller
