This dataset improves noise management precision by identifying and classifying open-office noise.

Train and evaluate RAG systems on flood emergency procedures with question-answer records grounded in source passages and supporting evidence.

Diagnose possible triggers of new convection from satellite time-series clues, with evidence and uncertainty clearly identified.

Explore four categories of waste in everyday kitchen scenes with annotations for training and evaluating recognition and classification models.

This dataset records the full video of a desert convoy crossing, supporting environmental monitoring and autonomous driving research.

Fine-grained annotations of change, negation, and uncertainty in satellite reports help improve model understanding and structured outputs.

This dataset focuses on detecting personnel working at heights in scaffold areas.

Build and evaluate retrieval and answering systems with evidence-linked questions grounded in extreme heat emergency plans.

Connect sea-level projection scenarios and quantitative details with source evidence to build more trustworthy retrieval-augmented question answering systems.

This dataset contains high-definition videos for analyzing and optimizing watering operations.

Train models to explain how sea-level changes affect coastal inundation risk by comparing baseline and hypothetical scenarios.

Use GIS coverage and boundary data for fog and low clouds to locate and analyze low-visibility weather regions.

Train models to explain how hypothetical precursor emission controls may affect ozone and why those effects depend on chemistry and environmental conditions.

A video dataset to enhance the performance of highway visibility estimation models under adverse weather conditions.

The Adverse Weather Road Image Classification Dataset focuses on enhancing the performance of intelligent transportation systems under complex meteorological conditions.

Train models to trace complex counterfactual causes linking external nutrient reductions, sediment release, and delayed water-quality responses.

Train discharge outlet detection models and support automated environmental inspections with images and location annotations.

Coral Reef Bleaching Disaster and Ecological Value Loss Prediction Image Dataset, aiding environmental protection and disaster analysis.

Train models to reason cautiously about changes in heavy metal contamination risks and migration under different remediation options and conditions.

Build traceable planning and compliance Q&A by locating key provisions in floodplain policies and flood protection regulations.