In today’s increasingly digitalized world, wireless protocols enable seamless data transmission across devices, supporting applications in telecommunications, healthcare, transportation, and more. However, the rapid advancement of wireless technologies has introduced new challenges, particularly in areas such as network security, device interoperability, and network reliability. Therefore, it has become crucial to develop a system capable of detecting and monitoring wireless Radio-Frequency (RF) communications, providing operators with the necessary data to address these issues. This paper presents a novel embedded system architecture for robust multi-channel, multi-protocol RF detection, classification, and localization, leveraging heterogeneous platforms and deep learning (DL)-based object detection. The proposed system addresses the challenge of broadband spectrum detection while ensuring high efficiency in both throughput and energy consumption. The DL models are trained on a custom RF dataset featuring a wide range of wireless protocols, taking into account low Signal-to-Noise Ratios (SNR) and Rayleigh fading scenarios. Our results demonstrate broadband capability, achieving an instantaneous bandwidth of 983.6 MHz while maintaining a good trade-off between accuracy (92%) and throughput (55 Million Samples Per Second (Msps)). Finally, the system is compact and lightweight, making it ideal for deployment in a portable embedded device.