Skip to content

Commit ac2194a

Browse files
noahbaculidjw-m
authored andcommitted
init perform ocr preparer and model docs
1 parent fe8a55d commit ac2194a

File tree

5 files changed

+176
-0
lines changed

5 files changed

+176
-0
lines changed
Lines changed: 47 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,47 @@
1+
---
2+
title: "NIM OCR"
3+
navTitle: "NIM OCR"
4+
description: "NIM OCR (Optical Character Recognition) is a model that extracts text from images on NVIDIA NIM."
5+
---
6+
7+
Model name: `nim_clip`
8+
9+
## About NVIDIA NIM Image OCR
10+
11+
The [NVIDIA NIM for Image OCR API](https://docs.nvidia.com/nim/ingestion/table-extraction/latest/overview.html) is an (Optical Character Recognition) microservice and interface that extracts text from images on NVIDIA NIM.
12+
13+
14+
## Supported aidb operations
15+
16+
* perform_ocr
17+
18+
## Supported models
19+
20+
### NVIDIA NIM for Image OCR
21+
22+
* [baidu/paddleocr](https://build.nvidia.com/baidu/paddleocr/modelcard) (default)
23+
24+
25+
## Creating the default model
26+
27+
```sql
28+
SELECT aidb.create_model(
29+
'my_paddle_ocr_model',
30+
'nim_paddle_ocr',
31+
credentials => '{"api_key": "__NVIDIA_NIM_API_KEY__" }'::JSONB
32+
);
33+
```
34+
35+
There is currently only one model supported by [NVIDIA NIM for Image OCR API](https://docs.nvidia.com/nim/ingestion/table-extraction/latest/support-matrix.html), the default `baidu/paddleocr`, so you don't need to specify the model in the configuration.
36+
37+
## Model configuration settings
38+
39+
The following configuration settings are available for CLIP models:
40+
41+
* `url` — The default is `https://ai.api.nvidia.com/v1/cv/baidu/paddleocr`.
42+
43+
## Model credentials
44+
45+
The following credentials are required if executing inside NVIDIA NGC:
46+
47+
* `api_key` — The NVIDIA Cloud API key to use for authentication.

0 commit comments

Comments
 (0)