Dataset: Transforming Network Traffic into Images: A Three-Branch Ensemble for Network Intrusion Detection in IoT Networks

Dataset accompanying "Transforming Network Traffic into Images: A Three-Branch Ensemble for Network Intrusion Detection in IoT Networks." Contains the raw packet captures, format-normalised captures, and per-flow split pcaps used to train and evaluate the three-branch ensemble described in the paper: 15 attack-family captures, a CICIDS-2017 benign subset and the authors' own normal_browsing capture. The generated packet-level and flow-level images (millions of files) are not included here since they are deterministically regenerated from this data using the code repository below. See the included README.md for the full directory structure, citation requirements for the underlying third-party sources (Rose et al.'s IEEE DataPort release, CICIDS-2017) and instructions for reproducing the paper's exact unified 70/15/15 split. Pipeline code: https://github.com/chriskarydis/Paper_Transforming_Network_Traffic_Into_Images