Custom Network Quantization Method for Lightweight CNN Acceleration on FPGAs

Table 3

Configuration of experimental environment.

Program	Detail
Hardware environment	CPU; GPU; Memory	Intel(R) Xeon(R) Gold 6148 CPU @ 2.40 GHz; Tesla V100 (graphics memory 16 GB); 100 GB
Software environment	Operating system; Deep learning framework; Acceleration library; Python; Datasets	Linux Ubuntu 16.04; PaddlePaddle 2.3.0; CUDA 12.0; Python 3.7.8; miniImageNet, CIFAR-10, OxFord 102 Flowers
Inference platform	Xilinx Zynq Ultrascale+MPSoC 3EG (XCZU3EG); Virtex UltraScale+XCVU13P (XCVU13P)