HANS: A Handwritten Answer Sheet Dataset for Noisy Hybrid Document Parsing
HANS targets the messy answer-sheet cases that current parsing benchmarks largely miss.
The paper introduces a dataset for handwritten student answer sheets containing text, math, hand-drawn tables, corrections, deletions, and strikethrough-style noise. The authors also propose NA-GOT, an end-to-end recognition framework with feature-level noise suppression and noise-aware decoding. Their experiments say existing methods struggle on HANS, while NA-GOT improves accuracy and stability. The dataset is planned for public release upon publication. ArXiv · AI/CL/LG's note
The paper introduces a dataset for handwritten student answer sheets containing text, math, hand-drawn tables, corrections, deletions, and strikethrough-style noise. The authors also propose NA-GOT, an end-to-end recognition framework with feature-level noise suppression and noise-aware decoding. Their experiments say existing methods struggle on HANS, while NA-GOT improves accuracy and stability. The dataset is planned for public release upon publication. ArXiv · AI/CL/LG's note
score 4