You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: articles/applied-ai-services/form-recognizer/faq.yml
+21-9Lines changed: 21 additions & 9 deletions
Original file line number
Diff line number
Diff line change
@@ -43,7 +43,7 @@ sections:
43
43
answer: |
44
44
45
45
Form Recognizer's deep-learning-based universal models support many languages that can extract multi-lingual text from your images and documents, including text lines with mixed languages.
46
-
46
+
47
47
Language support varies by Form Recognizer service functionality. See [language support](language-support.md) for a complete list of the handwritten and printed text supported by Form Recognizer.
48
48
49
49
- question: |
@@ -82,17 +82,17 @@ sections:
82
82
83
83
- question: |
84
84
What is the accuracy score and how is it calculated?
85
-
answer: |
86
-
The output of a `build` (v3.0) or `train` (v2.1) custom model operation includes the estimated accuracy score. This score represents the model's ability to accurately predict the labeled value on a visually similar document.
87
-
85
+
answer: |
86
+
The output of a `build` (v3.0) or `train` (v2.1) custom model operation includes the estimated accuracy score. This score represents the model's ability to accurately predict the labeled value on a visually similar document.
87
+
88
88
Accuracy is measured within a percentage value range between 0% (low) and 100% (high).
89
89
90
90
See [Interpret and improve accuracy and confidence scores](concept-accuracy-confidence.md#accuracy-scores)
91
91
92
92
- question: |
93
93
How can I improve accuracy scores?
94
94
answer: |
95
-
The accuracy of a model is influenced by variances in the visual structure of your documents.
95
+
The accuracy of a model is influenced by variances in the visual structure of your documents.
96
96
97
97
- Ensure that all variations of a document are included in the training dataset. Variations include different formats, for example, digital versus scanned PDFs.
98
98
@@ -177,15 +177,15 @@ sections:
177
177
| Supported Language | Azure SDK client-library|API reference |Supported API version|
What is the difference between Form Recognizer v3.0 and v2.1 and how do I migrate to the latest version?
186
186
answer: |
187
187
188
-
For improved usability, Form Recognizer v3.0 introduces a fully redesigned client library. To successfully use the latest Form Recognizer API features—version 2021-09-30-preview and newer—the most recent SDK is required and your application code must be updated to use the new clients.
188
+
For improved usability, Form Recognizer v3.0 introduces a fully redesigned client library. To successfully use the latest Form Recognizer API features—version 2021-09-30-preview and newer, the most recent SDK is required and your application code must be updated to use the new clients.
189
189
190
190
This table provides links to detailed instructions for migrating to the newest version of Form Recognizer:
191
191
@@ -253,6 +253,18 @@ sections:
253
253
254
254
Learn more about Form Recognizer [service quotas and limits](service-limits.md)
255
255
256
+
- question: |
257
+
How can I reduce request latency for my API calls?
258
+
answer: |
259
+
- Latency is defined as the amount of time it takes for an API server to handle and process an incoming request and deliver the outgoing response to the client. Occasional variability in latency and performance is inherent in any micro-service-based, stateless, asynchronous service that processes images and large documents at scale. While we're continuously scaling up the hardware and capacity and scaling capabilities, you may still see latency issues at run time.
260
+
261
+
- Try applying longer timeouts to account for random latency spikes. A timeout of 1 to 2 minutes should be sufficient to catch all responses. If not, try the call again after a short delay.
262
+
263
+
- Consider using [Proximity placement groups](/azure/virtual-machines/co-location) for workloads where low latency is a requirement.
264
+
265
+
- Enable a backup resource in another region to mitigate the latency impact.
266
+
267
+
256
268
- name: Custom models
257
269
questions:
258
270
- question: |
@@ -396,9 +408,9 @@ sections:
396
408
- question: |
397
409
Where can I find more solutions to my Azure Form Recognizer questions?
398
410
answer: |
399
-
[Microsoft Q & A](https://docs.microsoft.com/answers/topics/azure-form-recognizer.html) is the home for technical questions and answers at Microsoft. You can filter queries specifically related to Form Recognizer.
411
+
[Microsoft Q & A](https://docs.microsoft.com/answers/topics/azure-form-recognizer.html) is the home for technical questions and answers at Microsoft. You can filter queries specific to Form Recognizer.
400
412
401
413
- question: |
402
414
What should I do if specific text isn’t recognized or recognized incorrectly when labeling documents?
403
415
answer: |
404
-
We continually update and improve the Form Recognizer OCR model. Please reach out to the Form Recognizer team: [[email protected]](mailto:[email protected]). If possible, share a sample document with the issue highlighted.
416
+
We continually update and improve the Form Recognizer OCR model. You can reach out to the Form Recognizer team: [[email protected]](mailto:[email protected]). If possible, share a sample document with the issue highlighted.
0 commit comments