This article introduces a framework for rapid prototyping image processing
algorithms on Raspberry Pi and NVIDIA Jetson hardware platforms. The MATLAB
code for a chroma keying algorithm is provided together with images used in the
article and helper code used in deploying the MATLAB code to hardware.
Read the white paper here.
Murat Belge (2020). Code for Article "Run MATLAB Image Processing Algorithms on Raspberry Pi and NVIDIA Jetson" (https://www.mathworks.com/matlabcentral/fileexchange/66043-code-for-article-run-matlab-image-processing-algorithms-on-raspberry-pi-and-nvidia-jetson), MATLAB Central File Exchange. Retrieved .
Great example, thanks for the detailed example code. I'd like to bring up a small issue if I may. The comparison between the Raspberry Pi and the Jetson is misleading. The test on the Pi used double precision floating point computations, while the test on the Jetson used integers. A more interesting and direct comparison would involve first converting the initial algorithm to fixed point before generating the integer-based C code for the Pi. Or at the very least replacing double precision data with single precision. The Pi has native single-precision floating point hardware, but has to emulate double precision math which leads to a big performance hit. Looking forward to updated results if you care to dig a bit deeper.
No, no changes to the configuration is required. The generated code is the same for TK1 and TX boards.
To run the examples on Jetson TK1 is there any difference on the configurations to generate the code?
Added thumbnail image.
Inspired by: Run on Hardware