Cnn2rtl research data (v1.0)

Experimental data for Cnn2rtl, a CNN-to-VHDL generator. The deposit includes JSON/NPZ networks (MNIST Small/Medium/Large and Haddoc2 LeNet5, FaceDetect, CarType), OFAT synthesis manifests, quantization HW vs SW CSVs, Spearman/Pearson correlations, Haddoc2 comparison tables, generated VHDL for the three Haddoc2 designs, plot scripts, and the Vivado Tcl used to record Fmax, resources, and latency. The Python generator is not included. VHDL for the same models can be produced at https://cnn2rtl.itt-islab.com/ See README.md for the folder layout and how to check the numbers. MANIFEST.csv lists every file with SHA-256. License: data CC BY 4.0; scripts MIT.