126 - Testing a new application for third party monitoring of warehouses: Using artificial intelligence in Nigeria for USAID

Abstract Information: Decision makers use monitoring data to help course-correct toward target goals during the implementation of an activity/project. Monitoring projects can be costly, and often tradeoffs are made in the frequency of visits and number of sites selected for monitoring to reduce costs. This tradeoff usually leaves gaps in the information captured and limitations in analysis. Without monitoring, USAID investments have limited data about the efficacy of these activities. Additional challenges in monitoring exist when working in conflict environments often limiting the location, quantity, and quality of field visits. To address these gaps monitoring activities, teams need to identify new applications and methodologies to capture field visits more quickly, effectively, and with reduced risks to both respondents and team members. The Nigeria Monitoring Project (NMP) conducted a special study to test the use and benefits of a mobile application (app) in warehouse monitoring using an image segmentation algorithm for object detection and recognition. The sample included four implementing partner warehouses in the northeastern region of Nigeria. The team collected three rounds of data gathering and analyzing over 800 images. The findings indicate the artificially intelligence based application can yield correct counts of items, identify the type of item in an image, and provide additional information about the material of the product.