Bangla-Bayanno is an open-ended Visual Question Answering (VQA) dataset in Bangla, a widely used yet low-resource language in multimodal AI research.
While most existing VQA datasets are manually annotated with limited domains, query types, or answer formats, Bangla-Bayanno addresses these gaps by introducing a large-scale, high-quality dataset with diverse answer types. To minimize human errors and translation ambiguities, we applied a multilingual… See the full description on the dataset page:
https://huggingface.co/datasets/Remian9080/Bangla-Bayanno-Numeric.
