This paper addresses the detection and classification
of wideband Radio Frequency (RF) signals in the presence
of interference, particularly in low or negative Signal-to-Noise
Ratio (SNR) environments. The separation between the identifi-
cation and classification problems by employing two deep neural
networks (DNNs) to surpass the performance of traditional
methods is discussed. Additionally, it proposes the use of hybrid
Deep Learning (DL) architectures integrating information from
classical digital signal processing algorithms, a technique known
as contextual input. Finally, the implementation of a real-time
Field Programmable Gate Array-based (FPGA) detection and
classification system was proposed as a potential solution to
several current real-world problems.