Ibnuck/poster-schedule-information-extraction
Multimodal Visual-Text Dataset for Poster Schedule Information Extraction A ready-to-train Indonesian document AI dataset combining pixels, OCR tokens, spatial layout, and BIO entity labels. Overview What it is 127 Indonesian seminar and religious-study event posters with multimodal token-level annotations Primary task Schedule information extraction as token classification Modalities Image + text + 2D spatial layout Coordinates… See the full description on the dataset page: https://huggingface.co/datasets/Ibnuck/poster-schedule-information-extraction.
1115
