Two-stage RF pipeline for UAV/drone detection, parameter estimation, open-set model classification, and protocol recognition from wideband IQ. CNN P(drone) gate + ResNet/ArcFace classifier with open-set rejection, trained on RFUAV, Noisy Drone RF v2, and NIST wifi/bluetooth datasets. Kaggle-driven training.
deep-learning uav signal-processing cnn pytorch kaggle spectrogram resnet cnn-classification arcface wideband iq-data open-set-recognition signal-classification drone-detection rf-signal-processing
-
Updated
Aug 21, 2026 - Python