SurveySet for OCR and OMR-Based Survey Digitization

To support the development and evaluation of unified OCR–OMR models, we introduce SurveySet, a real-world dataset constructed from customer experience survey forms. SurveySet is designed to capture the diversity, informality, and structural variability found in physical survey documents. It includes both textual and visual input types, including handwritten digits, tick marks, crosses, and partially filled bubbles. This section describes the collection process, task-specific subsets, and annotation strategy. If you use the dataset for research purposes, please cite the orignal paper and dataset: Quiñones R, Cheekireddy S, Gultepe E. SurveyNet: A Unified Deep Learning Framework for OCR and OMR-Based Survey Digitization. Journal of Imaging. 2026; 12(4):175. https://doi.org/10.3390/jimaging12040175