1 個回答
- 最新
- 最多得票
- 最多評論
0
Textract would be your best bet - it's designed for handwriting recognition (among other things). Kendra is designed for language and meaning recognition. In super simplified terms, one is a convoluted neural network for images, the other is a recursive neural network for language; RNN works poorly for images, CNNs work poorly for language.
Textract does have limits - 3000 pages / 500MB for the pdf. Are you exceeding those limits?
已回答 2 年前
相關內容
- 已提問 6 個月前
- AWS 官方已更新 1 年前
- AWS 官方已更新 10 個月前