A number recognition system bulitd by RTL,and few IP. Can be used in Xilinx FPGA. tips:
1)There are detailed comments in the code。Download at ease
2)Using this project requires a Xilinx FPGA development board, ov7725 camera and HDMI display.
In the system, the OV7725 camera is first used to extract real-time image input classifier for recognition.
The input camera image is cut into an image size of 320*240, then converted into a grayscale,
and finally scaled into a grayscale size of 28*28 (the specific method is to use 8*8 matrix to cut the image,
Add the 64 pixels within each 8*8 matrix and assign them as new pixels) into the LeNet network.
LeNet network parameters are stored in the database (RAM)....Specific reports will be updated when available....
x7c325t Resource usage after deployment
test result
Letnet identification network section references the following design
https://github.com/suisuisi/FPGAandCNN/blob/main/images/Neura