Machine Learning Model to Predict Crowd Density for Public Event Accessibility

Inclusive Design Response Project (2 weeks)
June 2022

While attending a series of Toronto public events in the summer of 2022, the effect of crowd density on how able and disabled bodies could navigate these spaces, really stood out to me.

Events that were marked accessible, only indicated the provision of accessible features like washrooms, rather than being indicative of the event's overall accessibility.

To address these factors, I trained a machine learning model to identify how crowded a space was (using images), so people could decide if they would have the space to use their assistive mobility devices at public events.

In practice, the concept could work by scanning public social media pictures or live videos of a public event someone wanted to attend and indicate if it was too crowded or not (for their accessibility requirements).

Click on the video on the left to see the model in action, or refer to the keyframes below.

High density output read, in response to an image of a crowd

Medium density output read, in response to an image of a crowd

Low density output read, in response to an image of a crowd

Machine Learning Model to Predict Crowd Density for Public Event Accessibility

Inclusive Design Response Project (2 weeks) June 2022

Inclusive Design Response Project (2 weeks)
June 2022