crit

Company profile

Deep Vision Consulting is your expert in Computer Vision and 3D Vision at the edge.
The ready-to-use software assets in the Deep Vision Library are designed to easily turn your images, videos or 3D data into valuable high-level insights and carefully optimized for the execution over the most relevant embedded processors and AI accelerators. And in case your challenge requires an extra mile, our talented team of engineers and computer scientists is ready to create the customized solution for your specific case.

Specialties / excellences

Deep Vision Consulting covers the whole spectrum of computer vision projects, dealing with sensors (visible and IR spectrum, thermal, multispectral, 3D sensors: stereo, tof, lidar, profilometers), optics, 3D, computational geometry, machine learning and deep learning.

In addition Deep Vision Consulting is specialized in strategic and complementary topics to computer vision, such as

• optimizations for embedded systems (Arm CortexA and Neon extensions, Arm CortexM, neural accelerators from Nxp, Xilinx, Hailo, Analog Devices, Intel),
• robotics (cartesian and 6-Dof arms),
• simulations (Unity3D and proprietary simulators)
• augmented reality (Oculus rift/quest, HTC Vive Pro, Varjo XR-3).

Deep Vision Library

Deep Vision Library is a growing set of AI-based software IPs specifically designed to tackle challenges in Computer Vision and 3D Vision. At the edge, of course.

Indeed, each software IP in there is carefully optimized to be executed on embedded devices and neural accelerators.

The most recent software IPs introduced in the library are:

• Defect Visual Inspection: a compact, complete and cost-effective solution to grab images, train your AI model and identify your anomalies.
• AI for 3D reconstruction: create dense point clouds via stereo cameras with AI on embedded devices at full frame rate.
• 10 seconds Camera Calibration: automate and validate the calibration process via 6 degrees of freedom robot: make it quick, make it simple.
• Zero-dataset License Plate Recognition: no need to wait for training images: create your AI-based LPR just exploiting data synthesis.
• Milli-Watt AI: unleash the power of deep learning for computer vision with ultra-low power devices: battery-powered AI is real.