A fully pipelined, real-time, FPGA-based CNN classifier for Optical Character Recognition (OCR), built from the ground up — including convolution, ReLU, FC, quantization, and threshold-based decoding.