Enabling Fast and Accurate Crowdsourced Annotation for Elevation-Aware Flood Extent Mapping

Landon Dyken, Saugat Adhikari, Pravin Poudel, Steve Petruzza, Da Yan, Will Usher, Sidharth Kumar

In IEEE Pacific Visualization Conference (To Appear), 2025.

**Fig. 1:** An annotation made on a 6500 × 3000 pixel aerial image of flooding caused by Hurricane Matthew using only ten clicks of our topological segmentation tool. Pixels overlaid with blue have been labeled as dry and red as flooded. Note that areas labeled as flooded can appear green on the left, this is due to the tree canopy obscuring the flooded ground beneath. This tool guides annotation using the corresponding elevation data for a region, resulting in accurate labels. This can be seen by the highlighted areas where the tool annotated flooding around small dry features, but left them unlabeled. In our experimental user study, we show that this tool is more efficient than the state-of-the-art elevation-guided tool and can be used effectively by untrained participants

Abstract

Mapping the extent of flood events is a necessary and important aspect of disaster management. In recent years, deep learning methods have evolved as an effective tool to quickly label high-resolution imagery and provide necessary flood extent mappings. These methods, though, require large amounts of annotated training data to create models that are accurate and robust to new flooded imagery. In this work, we present FloodTrace, a web-based application that enables effective crowdsourcing of flooded region annotation for machine learning applications. To create this application, we conducted extensive interviews with domain experts to produce a set of formal requirements. Our work brings topological segmentation tools to the web and greatly improves annotation efficiency compared to the state-of-the-art. The user-friendliness of our solution allows researchers to outsource annotations to non-experts and utilize them to produce training data with equal quality to fully expert-labeled data. We conducted a user study to confirm our application’s effectiveness in which 266 graduate students annotated high-resolution aerial imagery from Hurricane Matthew in North Carolina. Experimental results show the efficiency benefits of our application for untrained users, with median annotation time less than half the state-of-the-art annotation method. In addition, using our application’s aggregation and correction framework, flood detection models trained on crowdsourced annotations were able to achieve performance equal to models trained on fully expert-labeled annotations, while requiring a fraction of the expert’s time.

Content

PDF

BibTeX

      @inproceedings{dyken_flood_ann25,
  author={Dyken, Landon and Adhikari, Saugat and Poudel, Pravin and Petruzza, Steve and Yan, Da and Usher, Will and Kumar, Sidharth},
  booktitle={IEEE Pacific Visualization Conference (To Appear)},
  title={Enabling Fast and Accurate Crowdsourced Annotation for Elevation-Aware Flood Extent Mapping},
  year={2025},
  }