Abstract-Floods are major natural disasters which cause deaths and material damages every year. Monitoring these events is crucial in order to reduce both the affected people and the economic losses. In this work we train and test three different Deep Learning segmentation algorithms to estimate the water area from river images, and compare their performances. We discuss the implementation of a novel data chain aimed to monitor river water levels by automatically process data collected from surveillance cameras, and to give alerts in case of high increases of the water level or flooding. We also create and openly publish the first image dataset for river water segmentation.

No abstract

No abstract

During natural disasters, situational awareness is needed to understand the situation and respond accordingly. A key need is assessing open roads for transporting emergency support to victims. This can be done via analysis of photos from affected areas with known location. This paper studies the problem of detecting blocked/open roads from photos during floods by applying a two-step approach based on classifiers: does the image have evidence of road? If it does, is the road passable or not? We propose a single double-ended neural network (NN) architecture which addresses both tasks simultaneously. Both problems are treated as a single class classification problem with the use of a compactness loss. The study was performed on a set of tweets, posted during flooding events, that contain (i) metadata and (ii) visual information. We studied the usefulness of each data source and the combination of both. Finally, we conducted a study of the performance gain from ensembling different networks. Through the experimental results, we prove that the proposed double-ended NN makes the model almost two times faster and the load on memory lighter while improving the results with respect to training two separate networks to solve each problem independently.

No abstract

No abstract

This paper proposes a novel method to optimize bandwidth usage for object detection in critical communication scenarios. We develop two operating models of active information seeking. The first model identifies promising regions in low resolution imagery and progressively requests higher resolution regions on which to perform recognition of higher semantic quality. The second model identifies promising regions in low resolution imagery while simultaneously predicting the approximate location of the object of higher semantic quality. From this general framework, we develop a car recognition system via identification of its license plate and evaluate the performance of both models on a car dataset that we introduce. Results are compared with traditional JPEG compression and demonstrate that our system saves up to one order of magnitude of bandwidth while sacrificing little in terms of recognition performance.

No abstract