Deep Learning Benchmarks and Datasets for Social Media Image Classification for Disaster Response

Image | General | Response

This dataset is a combination of DAD, CrisisMMD and DMD data for general-disaster damage severity assessment. It contains 34,896 images and is introduced for multiclass (ordinal) classification.

  • ML task type: Multiclass (ordinal) classification
  • Data Source: Social Media (Twitter, Google, Instagram)
  • Size: 34,896 Images
  • Timespan: N/A
  • Geographical Coverage: Global
  • Baseline Information
  • Evaluated on: ResNet18, ResNet50, ResNet101, AlexNet, VGG16, DenseNet, SqueezeNet, InceptionNet, MobileNet, EfficientNet
  • Metrics used: F1-score
  • Results as reported in original paper: F1-score: 0.758 (EfficientNet)

Firoj Alam, Ferda Ofli, Muhammad Imran, Tanvirul Alam, and Umair Qazi. Deep learning benchmarks and datasets for social media image classification for disaster response. In 2020 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining, pages 151–158. IEEE, 2020.