ECE.25 – DNGR Detection System

• Pablo Castro
• Sung Jun Choi Hong
• Eduardo Gijon
• Jason Jimenez

Gun crimes have become a source of senseless violence and tragic deaths in the city of Chicago and similar urban areas. In a city like Chicago, many of these gun crimes have a high chance of going unsolved with police presence being unable to assist every victim that becomes unfortunately involved. Current technologies in use by the city tend to report false positives tying up the already small police force and worsening relations with police while perpetuating discrimination due to the placement of Shotspotter in minority neighborhoods. DNGR is an acoustic gunshot detection system which will utilize pattern recognition and machine learning software in order to accurately identify gunshots with up to a 95% detection rate around the vicinity of a homeowner’s property and alert them of events through a smart app. The app will provide owners a timestamped notification of when the event occurred and give the option to provide the data to police authorities. The system will also utilize the ability to connect with smart home APIs such as Google Assistant and Amazon Alexa in order to notify and gather data on time logs of gunshots around their proximity while also pushing notifications to any other connected smart devices on a home network. In dealing with this data and the recording process, user privacy will be maintained through the system per user surveys and feedback to maintain peace of mind. As a team, we have implemented a trained neural network to filter gunshots through a voice bonnet attached to a Raspberry Pi 4 which continuously detects, records and sends notification data to a graphic user interface application while also supporting Google Assistant commands. LEDs have been programmed to provide a visual cue of the system status.