We were inspired because we all had an interest in learning machine learning, hardware and concerned about the inability to respond accurately to crime. Our firm, InsomniHacks, has created a crime detection system that will be placed in high crime locations. Our device uses a microphone and camera to monitor suspicious sounds and events, and our machine learning application constantly monitors for and determines unsafe audio, by cross-referencing our customized database of urban sounds. Our algorithm will determine the degree of danger presented by the audio, and trigger appropriate responses to authorities, through text API service Twilio. We used PyAudioAnalysis as our machine learning library to classify screams and be able to detect them from other sounds. Screams are associated with violence and victims of various crimes. Accurately, training a machine learning model and having a large enough data set. We are proud of learning two new skills at HTN. We learned how to incorporate our own machine learning algorithm and how to use various hardware's to interact with the client such as Python and sms. Firefly would like to incorporate a 360 degree camera and mic that are able to detect distance and direction of sound. In addition, to live streaming media to police.


Placed 2nd place for Situational Awareness Challenge for Canadian Special Operations Forces Command (CANSOFCOM)

Share this project: