[ Real-time operation on ARM CPU ]

Uncanny Vision's software achieves excellent real-time performance on the ARM Cortex-A series, enabling it to solve the following challenges in computer vision:

• Communication band

To use computer vision software that runs on the cloud, it is necessary to continuously send large amounts of video data from the camera to the cloud.

• speed performance

The processing takes too long to operate on an embedded system, and it is difficult to perform the expected detection and recognition in real time.

• Built-in

Using a GPU allows computer vision to be done locally (without connecting to the cloud), but it consumes more power and costs more.

[Application example]

It is a technology that can be applied in various fields such as machine vision, in-vehicle cameras, surveillance systems, gesture recognition, drones, and robotics.

• Face, pedestrian, animal detection
• object, scene recognition
• object tracking
• Human Posture Recognition and Action Prediction

[Can be used in any environment]

In addition to Linux, we have a proven track record on all embedded platforms such as iOS, Android, QNX, and INTEGRITY. Also, it can be used not only in the ARM Cortex-A series, but also in the Intel x86 environment.

[ Uncanny CV ]

A fast computer vision library optimized for embedded use.
It has over 70 image processing algorithms and functions.

[ Uncanny DL ]

A fast deep learning library optimized for embedded use.
Higher accuracy can be achieved by using it in combination with UncannyCV.

[ Development Service ]

Development of computer vision algorithms, acceleration, FPGA
We also provide support such as porting and learning work on behalf of you.