Barcode, OCR, and Simulated Responses
We recently implemented some improvements to the API:
- A
/barode-onlyspeedup - An improved OCR algorithm
🏎️ Refactored /barcode-only
/barcode-onlyWe refactored our code to improve the /barcode-only endpoint response time. You should see a significant reduction in latency per response. This speedup does not affect the /barcode-only request or response schemas. Enjoy life in the fast lane!
👀 New OCR algorithm
There's a new OCR algorithm at work behind the scenes. It handles all North American IDs (a document_type of na_dl). Now, for all na_dl requests, if your signals array includes ocr_scan or ocr_match, we use the new algorithm. It's faster and more accurate. Plus, we're able to handle poor images with better precision. The response schema is unchanged.
Other changes
Return user_defined if configured
user_defined if configured- The
user_definedproperty is typically used for your own internal tracking requirements. You can submit auser_definedobject in the/startrequest. We updated our backend systems to ensure thatuser_definedis returned in the/startresponse for all properly configured accounts.
Nationality returned
- We fixed an issue for
document_typeofpassportorotherwherenationalitycould be returned asnullin theOCRorextendedIdresponse objects.
Improved Try It! button
When you click the Try It! button on the API explorer, you get simulated results in the Response box. Simulated results provide randomized (but realistic) data and follow the same response schemas as production endpoints. It's a great way to quickly become familiar with each of our endpoints and see how they respond to various requests-- no code required!
To learn more, see Try It Without Code.
The Try It! button returns better, more accurate simulated results:
- We fixed the
/submit-email-addressand the/get-statusendpoints. - We made numerous corrections to more closely align the simulated results with production results.
- We added more simulated result objects for the
/get-resultsendpoint.
So now the simulated results are pretty accurate. But discrepancies are still possible. Feel free to use the Try It! button for exploring and learning. However, we humbly ask that you do not write production code against simulated results. 🙏 Save that for your staging environment. Learn more here: Sandbox and Staging.
