AI for Urban Resilience in Africa: Pest Detection in Cabo Verde Using Convolutional Neural Networks

Rapid urbanization in Africa is placing increasing pressure on food systems and aggravating challenges related to food security, particularly in small island developing states that face specific vulnerabilities. In Cabo Verde, agricultural production is constrained mainly by water scarcity and pest proliferation, both of which reduce productivity and compromise food supply in urban and rural areas. In this context, artificial intelligence (AI) emerges as a strategic tool to support pest identification and decision‐making processes, thereby enabling timely interventions and contributing to the resilience of agri‐food systems. This study presents the development of an AI‐based system for pest identification using convolutional neural networks and transfer learning with the MobileNetV2 architecture. The model was trained and tested on 1,028 images representing four of the most recurrent pests in Cabo Verdean agriculture— Spodoptera frugiperda (518), Nezara viridula (252), Acherontia atropos (120), and Chrysodeixis chalcites (138)—collected from open sources and local farming fields. Data augmentation and class weighting were applied to mitigate dataset imbalance, resulting in a test accuracy of 94.17% and strong generalization capacity. The trained model was integrated into a web application that enables users to upload pest images and receive real‐time identification along with practical recommendations for biological and chemical control. The study demonstrates that the success of AI in Africa relies on its alignment with local contexts and its ability to deliver simple, accessible, and low‐cost solutions that directly support community livelihoods. Future integration of a citizen science component could further enhance pest mapping and strengthen collective responses to food security challenges.