This document provides an overview of the Flood-related Multimedia Task, which aimed to analyze social media posts from Twitter to determine their relevance to actual flooding incidents in northeast Italy. The task involved classifying tweets containing Italian keywords related to floods as either relevant or not relevant to flooding in the target area. 5 teams participated and submitted runs using text, image, or combined features. The best performance was from undersampling and combining three artificial neural networks, achieving a maximum F1-score of 0.5405 for classifying the tweets, though on average performance was low, indicating this was a challenging task.