Detecting Public safety risks using sound cameras

• Creating situational awareness using sound
• Robust localization and classification of sound sources
• Real-time
• Privacy robust solution using only metadata and short audio clips as output
• Multiple classifiers can be used to trigger events (Aggression, car alarms, gunshots, vehicle detection, drones).
• Integration with multiple platforms through API

Input(s):

• Timestamped audio streams from multiple sound cameras
• Classification of aggressive sound

Main feature(s):

• Sound cameras listen to the specific surrounding and are able to detect and classify aggressive behavior using sound classification software
• Detection and event alarms can be used to inform police or other public authorities through a user platform (e.g. Unity or Esri platform)

Output(s):

• Metadata information
• Short (max 2 minutes) pre- and post alarm audio recording

• Integration through API
• SW constraints: no
• HW constraints: NVIDIA GPU, 8 GB GPU RAM

Police, surveillance and security operators, municipalities, traffic monitoring.

• Hardware (sound cameras)
• Licensing (API and classification software)