Unattended Item Detection

Problem Statement

Identifying Unattended items on real-time streaming video.

What it does

The solution would assist in the Security of stations, roads and generate an alarm upon detection of abandoned / unattended items. Such items could potentially be instruments of harm or mass destruction targeting a public location or alternately could be cases of misplaced or lost items/theft.

Challenges we ran into

• Real-time distance calculation with auto camera calibration.

Accomplishments that we're proud of

• Significantly enhance security in public transit locations
• With falling prices of CCTV cameras, the solution can become ubiquitous
• Reduce need for human officers to continuously monitor CCTV streams
• No human error due to fatigue, lack of sleep, or other reasons

What we learned

Object Detection with Computer Vision

What's next for Unattended Item Detection

• Handling of Occlusion
• Illumination changes
• Low-quality recording
• Owner Detection of Unattended Item

Built With

• computervision
• flask
• imageprocessing
• machine-learning
• python