Swirled images of CFD simulations with different rotation angles

This dataset contains high-resolution CFD simulation images of airflow patterns in naturally ventilated barns. Each image shows 2D cross-sections of flow fields, such as velocity distributions, under various environmental and structural conditions. Swirl transformations were applied around salient flow regions at 30° rotation steps, from 0° to 360°, to introduce controlled rotational distortions. All images are formatted as 8-bit grayscale PNGs (1380×1800 px). This dataset is used to evaluate image similarity metrics and train machine learning models for predicting airflow perturbations. For more details, please see the relevant article: "Assessing Airflow Perturbations in Naturally Ventilated Barns" (2025). Images for the case study based on real-world sensor measurements are also included in this repository (see the respective README).