The problem Social Distancing Detector solves

1. Airborne disease spread - The project aims to reduce disease spread by finding congregation hot spots, allowing for better utilization of space and determination of proper protocols.
2. An IWFM research shows that almost 41% employees don’t plan on returning to the office until the workplace is made COVID safe - our system ensures that social distance protocols are maintained in the workplace so employee saftey is guarenteed and employee satisfaction is maintained.
3. Provides valuable analytics and insights on crowd congregation. Helps inform authorities to identify likely crowded areas and build policies accordingly.
4. In retail and healthcare sectors, helps detect if people are following social distance guidelines.
5. Provides social distance monitoring, acting as a base for a system to help ensure social distance protocols are maintained in public places, retail/hospitals, workplace and in the streets.

Challenges we ran into

1. The model didn't have good perfomance on the perspective used by monitoring cameras, but we overcame that by using Triangle Similarity Calibration, which mapped the pixel distance with the real distance metric.
2. The major problem which limited us was that of time and hardware. We used a pre-trained Object Detection model and didn't train one on our own which limited us in terms of flexibility.