PerceptionDataset \u2013 Synthetic Radar Perception Dataset (Random Burst Targets)

This dataset contains a fully synthetic multimodal radar perception dataset designed for evaluating perception-aware learning in cognitive radar networks with limited sensing resources. It includes (i) STFT-based 2D spectrogram features for CNN models, and (ii) complex I\/Q baseband radar returns for sequential deep learning models. Each instance represents a radar burst labeled as useful or non-useful, enabling binary perception classification tasks. The dataset supports research in cognitive radar, spectrum-efficient sensing, AI-based radar perception, and reproducible benchmarking.