This project involves loading a pre-trained state-of-the-art CNN to classify radio signals (with inputs as spectogram images) into Squiggle, Noises, Narrowband, etc. Furthermore, we apply spectogram augmentation using time & frequency masking. This could practically be used in contexts involving military, security, telecommunications, and space technology.
EfficientNet-B0 consists of seven blocks which are shown in different colours. The basic building block of EfficientNet-B0 is a mobile inverted bottleneck convolution (MBConv), while each MBConv block is shown with the corresponding kernel filter size.
Dhameliya, P. (n.d). Classify Radio Signals with PyTorch [MOOC]. Coursera. https://www.coursera.org/learn/classify-radio-signals-with-pytorch
Zhou, Aihua & Ma, Yujun & Ji, Wanting & Zong, Ming & Yang, Pei & Wu, Min & Liu, Mingzhe. (2022). Multi-head attention-based two-stream EfficientNet for action recognition. Multimedia Systems. 29. 1-12. 10.1007/s00530-022-00961-3.
PyTorch. (n.d). PyTorch documentation. https://docs.pytorch.org/docs
torchvision. (n.d). PyTorch's torchvision library documentation. https://docs.pytorch.org/vision/
Prefix.dev. (n.d.). Pixi documentation. https://pixi.prefix.dev/latest/