CUVOS

MicroRaptor

Enhancement on SWaP-constrained platforms

EagleEye models how retinal cells behave and interact. MicroRaptor reaches further back, to photoreception that has nothing to do with sight, the kind simple animals use to read light without forming an image.

MicroRaptor costs less to implement. In difficult lighting it holds its own against the more complex approach, making it an option if your power budget is constrained.

A low-light example

Below is the output of MicroRaptor running on a RaspberryPi. Original video is on the left. Both images and the metrics were generated in real time.

Deployment options

We have run MicroRaptor on ARM processors and on Xilinx Artix 7 FPGAs. Some relevant stats are shown below. Developing lower-SWaP algorithms based on this architecture are part of our active R&D activities.

On Xilinx Artix 7

Form
IP block
Power
<200 mW
FPGA resources
<5%

Microprocessor build

Processor
4-core ARM
Capture and processing1
2.5 W
System total1
6.5 W

Dark-room test

MSE
109.67
PSNR
27.73 dB
SSIM
0.151
Processing time
6.89 ms

1 Measured on a 4-core ARM board. The 2.5 W covers acquisition as well as the MicroRaptor processing; 6.5 W is what the complete system draws with it running.

Try it on your own footage

Contact us to arrange a demonstration based on a video you provide. We have a range of mechanisms for facilitating this, depending on the nature and sensitivity of your data.

Contact