Geography

Make a Gift

RESEARCH HIGHLIGHT

Perception bias: Deciphering a Mismatch Between Urban Crime and Perception of Safety

PURPOSE

In his research, Dr. Yujie Hu explores how to answer one question—are places looking safe really safe or vice versa?

DR. YUJIE HU

Step 1: Evaluate People’s Risk Perception

Utilize deep learning to analyze 384,180 Google Street View images and evaluate people’s risk perception of streets and neighborhoods in Houston. Below are sample images that demonstrate perceived high and low safety scores.

ROW A: IMAGES WITH HIGH SAFETY SCORES

ROW B: IMAGES WITH LOW SAFETY SCORES

Step 2: Compare Perception with Actual Risk Level

The derived risk perception is then compared with the actual risk level measured by crime rates to reflect the so-called perception bias. Results capture the level of mismatch between people’s visually perceived safety and the actual crime rates. This measure provides scalable guidance in deciphering the relationship between the built environment and crime. This research can enhance our capacity to design urban management strategies that prevent the emergence of extreme “perception bias”.