Quick Summary
- ▸ The project consisted of analyzing a real-time video stream and identifying events and the presence of humans in demarcated areas.
- Delivered a system that identifies events and the presence of people in demarcated areas in real time.
- Delivered multi-channel alert reporting via phone calls, text messages, and email upon intrusion detection.
Overview
The system was networked with a set of up to 64 cameras on a LAN (Local Area Network), also using internal camera event detection mechanisms or polling for cameras that did not support these events. Video streams were extracted and passed through the OpenCV computer vision mechanism to detect intruders using a classification algorithm tuned internally for the project. When an intruder was detected, alerts were generated to up to five contacts using Twilio and SMS, as well as emails and pre-recorded calls. In this case study, we delve into the intricacies of our innovative solution that integrates artificial intelligence into the video surveillance area. Our platform stands out as the demand for advanced security measures intensifies, combining state-of-the-art computer vision, machine learning algorithms, and real-time analytics.
Industries
Services Provided
- Web Development
- Custom Software Development
- AI Development
Approach
The technologies used in this project are Python, JavaScript, YOLOv3, and OpenCV. This project lasted two years, with one senior and one semi-senior developer.
Results
- Delivered a system that identifies events and the presence of people in demarcated areas in real time.
- Delivered multi-channel alert reporting via phone calls, text messages, and email upon intrusion detection.
- Integrated with home automation software to enable coordinated security responses.
- Supported a network of up to 64 cameras over LAN with mixed event-detection and polling capabilities.
