Computer Vision and Search
Pre-built vision and search services.
| Service | Input | Job | Key distinction |
|---|---|---|---|
| Amazon Rekognition | Images/video | Objects, faces, moderation | General vision; incidental text only |
| Amazon Textract | Documents/PDFs | Extract forms, tables, key-values | Preserves structure, not just OCR |
| Amazon Kendra | Documents | Natural-language enterprise search | Q&A out of the box, no relevance tuning |
- Rekognition (photos/video vision) vs Textract (structured document extraction) — both “read”, different jobs.
- Kendra (NL question answering) vs Amazon OpenSearch Service (keyword/log search you tune).