Sign in to view source links and access this dataset
Description
1,395 real-world disaster images and 4,405 expert-curated question–answer pairs covering floods, wildfires, and earthquakes. The dataset includes binary, multiple-choice, and open-ended questions for evaluating Vision-Language Models. It was created by QCRI and last updated in May 2026.
Use Cases
Benchmarking VLM performance on disaster scene understanding based on expert-curated questions.
Training models for binary (Yes/No) classification of disaster scene attributes.
Evaluating model reasoning on multiple-choice questions about disaster impact.
Developing open-ended visual question answering systems for disaster response scenarios.
Strengths
Contains 4,405 expert-curated question–answer pairs, providing a substantial benchmark.
Includes 1,395 real-world images covering three disaster types (floods, wildfires, earthquakes).
Features three distinct question types (binary, multiple-choice, open-ended) for comprehensive evaluation.
Limitations
Column-level documentation is absent; field semantics must be inferred after download.
Row count is unknown, which may limit suitability assessment.
Freshness should be verified as the last update timestamp is in the future (2026-05-20).
Provenance
Source
QCRI
Collection Method
Expert-curated collection of real-world disaster images and questions.
Freshness
Last updated 2026-05-20 07:52:09
License is unknown; terms of use must be verified before application.