Introduction
Leveraging advanced technology for improved accuracy
Video surveillance is now a part of modern life and helps law enforcement make better decisions. Yet, even with vigilant human operators, errors can occur due to limited attention spans and difficulty in tracking multiple events.
To improve efficiency, we need more than human processing. We need advanced technology.
Challenge
Overcoming the limitations of traditional surveillance
Our client, a drone manufacturer in India, wanted to improve the technology of their drones to capture and store video footage.
They needed a way to analyze the footage automatically, especially at night when traditional thermal imagery doesn’t work well with visual images. Human operators were doing the analysis, which slowed the process and wasn’t always accurate.
Solution
Our step-by-step approach to creating an advanced IVA platform
Fractal proposed an Image and Video Analytics (IVA) platform using advanced deep learning techniques, which could automate the analysis and improve precision and speed, making video surveillance more effective.
Steps
What We Provided:
Fractal’s solution: advanced threat detection
By the end of the third phase, we had developed and extensively tested a scalable system. This highly accurate and robust threat detection engine outperformed human operators in various environments.
The engine could perform complicated tasks, including the ability to:
● Trail a detected person
● Generate alerts on social distancing protocols
● Perform in challenging environments (e.g., high altitude, crowded spaces, and forests)
Furthermore:
● A social distancing compliance protocol was developed to assist with COVID-19 public safety.
● A daytime person detection engine was developed and tested in urban and rural environments.
● The daytime detection engine was extended to include solutions to the challenging problem of person detection at nighttime.
The results
The future of security