Requirements and limitations for a form processing model
Afrikaans, Albanian, Asturian, Basque, Bislama, Breton, Catalan, Cebuano, Chamorro, Chinese (Simplified), Chinese (Traditional), Cornish, Corsican, Crimean Tatar (Latin), Czech, Danish, Dutch, English, Estonian, Fijian, Filipino, Finnish, French, Friulian, Galician, German, Gilbertese, Greenlandic, Haitian Creole, Hani, Hmong Daw (Latin), Hungarian, Indonesian, Interlingua, Inuktitut (Latin), Irish, Italian, Japanese, Javanese, K’iche’, Kabuverdianu, Kachin (Latin), Kara-Kalpak, Kashubian, Khasi, Korean, Kurdish (latin), Luxembourgish, Malay (Latin), Manx, Neapolitan, Norwegian, Occitan, Polish, Portuguese, Romansh, Scots, Scottish Gaelic, Slovenian, Spanish, Swahili (Latin), Swedish, Tatar (Latin), Tetum, Turkish, Upper Sorbian, Uzbek (Latin), Volapük, Walser, Western Frisian, Yucatec Maya, Zhuang, Zulu.
Form processing works on input documents that meet the following requirements:
- JPG, PNG, or PDF format (text or scanned). Text-embedded PDFs are better, because there won't be any errors in character extraction and location.
- TIFF files cannot be used for training. You will need to use documents in PDF, JPG or PNG format to train a model. Once the model has been trained, it can extract data from TIFF files when the model is used in a Power Automate cloud flow.
- If your PDFs are password-locked, you must remove the lock before submitting them.
- The combined file size of the documents used for training per collection must not exceed 50 MB.
- For images, dimensions must be between 50 × 50 and 10,000 × 10,000 pixels.
- For PDF files, dimensions must be at most 17 x 17 inches, corresponding to Legal or A3 paper sizes and smaller.
- If scanned from paper documents, scans should be high-quality images.
- You can create up to 200 collections per model.
AI Builder doesn't currently support signatures as a type of form processing input data.
Learn how to improve the performance of form processing models.