A deep learning approach to water point detection and mapping using street-level imagery

Households in developing countries often rely on alternative shared water sources that exist outside of the datasets of public service providers. This poses a significant challenge to accurately measuring the number of households outside the public service system that use a safe and accessible water source. This article proposes a novel deep learning approach that utilizes a convolutional neural network to detect water points in street-level imagery from Google Street View. Using a case study of the Agege local government area in Lagos, Nigeria, the model detected 36 previously unregistered water points with 94.7% precision. HIGHLIGHTS The article presents a deep learning model that extracts street-level imagery from Google Street View and uses a YOLOv5 object detection model to detect shared water points using street-level imagery. The model detects 36 previously unregistered water points with 94.7% precision using a pilot study of Agege LGA in Lagos, Nigeria. Model performance is evaluated in terms of inference speed, cost, and scalability.