Vehicles had to be located and their movements tracked but without video cameras (due to social constraints) and large volumes of still cameras (expensive).
Using recorded stereo audio signals (from microphones placed about 5m apart on a highway), Alumnus teams designed and implemented a solution to determine vehicle attributes on a street: Vehicle Size (cars or mid-sized trucks or buses/trolleys), Direction of movement (left-to-right or right-to-left).
This could overcome the unavailability of video surveillance. Recorded raw WAV files of one-two minutes durations (from microphones placed about 5m apart) were processed to determine audio energy to indicate vehicle size and some clever frequency domain computations (from short-term FFT obtained using GNU Octave) like convolution and correlation and filtering were carried out to determine vehicle direction.
Real-time sound reckoning, Short Time Fourier Transform (STFT), Generalized Cross-Correlation with Phase Tracking (GCC-PHAT), GNU Octave, Python