fpga-based face detection and recognition

[kinupusan]

I have already working matlab codes for both detection and recognition. the problem is i don't know where to start in implementing it to fpga. i know little of verilog but i can learn fast. can somebody help?:?: thanks in advance.

 

[TrickyDicky]

I suggest learning more verilog, and specifically learning verilog for synthesis.

 

[kinupusan]

thank you for your suggestion. I already have a working matlab code and I plan to implement it in an Altera DE2 FPGA board. the first problem would be transferring the image file from PC to FPGA? any suggestions?

 

[aruipksni]

thank you for your suggestion. I already have a working matlab code and I plan to implement it in an Altera DE2 FPGA board. the first problem would be transferring the image file from PC to FPGA? any suggestions?

Try With Altera System Console :

https://www.altera.com/literature/hb/qts/qts_qii53028.pdf

and if you interest in matlab interfacing have also a look at :

https://www.altera.com/technology/dsp/system-in-the-loop/fpga-verification-matlab.html

 

[sreevenkjan]

you can store the image in the block ram of the fpga....it again depends on the memory size of the image and also the memory available in the fpga..

 

[guy.lemieux]

Face detection is easily done in software using a soft-core processor. Some attempts at hardware versions have been described by researchers -- the most recent work I know is by Prof. Jonathan Rose, which improves earlier work by Prof. Ryan Kastner.

Converting a matlab algorithm to FPGA can be assisted by tools like DSP Builder (from Altera), System Generator for DSP (Xilinx) and HDLcoder (from Mathworks).

If MicroBlaze or Nios are too slow for the image processing, you can look for vector accelerators. One such accelerator is made by VectorBlox. They have a face detection demonstration that is ~10x faster than the ARM processor on Zynq devices.