AI to Support Citizenship

Introduction

In an era of rapid technological advances, the relationship between Artificial Intelligence (AI) and everyday life is continuously expanding its horizons. An obvious example of this synergy is the emergence of AI-based applications that enable citizens to actively engage in improving their communities. One such solution is the prototype of an AI-based mobile application designed to allow citizens to report and address urban problems they encounter through photographs.

Proposed Solution

The traditional process of identifying and solving urban problems, such as potholes, damaged infrastructure or water infiltration, often relies on manual inspection or occasional reporting, causing delays in problem solving. The app proposed by Ennova Research aims to revolutionise this paradigm by putting the power to report problems directly in the hands of community members. The app promises to speed up problem detection, increase civic participation and ultimately contribute to the creation of cleaner, safer and more efficient urban environments.

The pipeline developed for the app is as follows:

1. Capturing the problem: The first step of our solution is very simple. Users encounter an urban problem, such as a pothole or water infiltration, and decide to take action. They open the mobile app and use their smartphone camera to take a picture of the problem. This picture starts the process of solving the problem.
2. Sending the Image to the Model: Once the user uploads the photo, the app automatically sends the image to the AI model, deployed on the Google Cloud Platform. The model processes the image, extracting relevant information and identifying the nature of the problem. This almost instantaneous analysis is at the heart of the mobile app.
3. Inference Verification: Building trust between the AI and the user is critical to the success of our solution. To achieve this, the app presents the AI-generated problem classification to the user for verification. A simple click or tap allows the user to validate the inference of the model, confirming whether it correctly identifies the problem. This human validation ensures the reliability and accuracy of the problem classification.
4. Archiving of Labelled Images: Once the user’s confirmation is received, the labelled image, together with the associated problem category, is securely stored in a dedicated database. This database serves two purposes. First, it keeps a record of reported problems for reference and monitoring. Second, it constitutes a valuable dataset that is instrumental in the continuous training and improvement of the AI model
5. Continuous Model Improvement: Periodically, the AI model is re-trained using this data, allowing it to improve its problem recognition capabilities over time. As the model becomes more adept at identifying urban problems, the accuracy and efficiency of the entire reporting process increases, ultimately benefiting both users and the community at large

Results

The architectural choices proved to be in line with user expectations. A high retention rate, as well as an increase in the number of images stored in the database, measured the success of the initiative towards the community. The app developed by Ennova Research is an example of how AI technology can best serve people’s needs when a user-centred approach is established early in the design and development process.