Simplifying reporting of street issues using image classification
The challenge
The 'Report it' service allows residents, businesses, and visitors to report street issues like fly-tipping, graffiti, animal fouling and street noise. However, the process required users to fill out forms and manually select the issue type, making it time-consuming and prone to errors. This discouraged reporting and led to delays in resolution when issues were misclassified.
The solution
The 'Report it' Assistant (RIA) uses AI to automatically identify the type of issue using a combination of image classification (on user-uploaded photos) and natural language understanding. This removes the need for users to manually select categories, speeding up the reporting process. It also ensures accurate onward triage and action, improving the overall resident experience.
The results
The AI correctly identified the report category in 86% of cases, leading to a 38% reduction in time spent selecting the right category and a 20% drop in the number of steps required. Overall, the average time to report an issue fell from 3.65 minutes to 1.37 minutes.
Learnings / insights
Incident triage is a feasible and effective opportunity space for machine-learning application.
'Report It' Assistant
Simplifying reporting of street issues using image classification
The challenge
The 'Report it' service allows residents, businesses, and visitors to report street issues like fly-tipping, graffiti, animal fouling and street noise. However, the process required users to fill out forms and manually select the issue type, making it time-consuming and prone to errors. This discouraged reporting and led to delays in resolution when issues were misclassified.
The solution
The 'Report it' Assistant (RIA) uses AI to automatically identify the type of issue using a combination of image classification (on user-uploaded photos) and natural language understanding. This removes the need for users to manually select categories, speeding up the reporting process. It also ensures accurate onward triage and action, improving the overall resident experience.
The results
The AI correctly identified the report category in 86% of cases, leading to a 38% reduction in time spent selecting the right category and a 20% drop in the number of steps required. Overall, the average time to report an issue fell from 3.65 minutes to 1.37 minutes.
Learnings / insights
Details
Links