A system for detecting, classifying and tracking unmanned aerial vehicles (UAVs) comprising: at least one microphone array arranged to provide audio data; at least one camera arranged to provide video data; and at least one processor arranged to generate a spatial detection probability map comprising a set of spatial cells. The processor assigns a probability score to each cell as a function of: an audio analysis score generated by comparing audio data to a library of audio signatures; an audio intensity score generated by evaluating a power of at least a portion of a spectrum of the audio data; and a video analysis score generated by using an image processing algorithm to analyse the video data. The system is arranged to indicate that a UAV has been detected in one or more spatial cells if the associated probability score exceeds a predetermined detection threshold.