AI-Surveillance-System

The problem AI-Surveillance-System solves In an era of increasing public safety concerns, traditional surveillance systems face critical limitations in proactively identifying potential threats. Current CCTV infrastructure primarily serves as passive recording devices, requiring human operators to manually monitor countless hours of footage, leading to delayed response times and missed threats. The system addresses the urgent need for real-time weapon detection and suspicious behavior analysis by leveraging artificial intelligence to automatically identify firearms, knives, and other dangerous objects in live video feeds. By integrating emotion recognition and behavioral analysis, it provides early warning capabilities that enable law enforcement to respond proactively rather than reactively, potentially preventing violent incidents before they escalate and creating safer public environments through intelligent, automated vigilance

Challenges we ran into When we started this project, our biggest reality check was realizing we had virtually NO usable data. We thought we could just download a dataset and start building, but we quickly discovered that public weapon detection datasets were either: too small, poorly labeled, or completely unrealistic for real surveillance scenarios

Built With

• firebase
• flask/fastapi-(backend)
• html5
• javascript
• opencv
• python
• react-or-html/css/javascript-(frontend)
• sqlite/mongodb-(database)
• tensorflow/pytorch
• yolo-(for-object-detection)