Most high-quality images yield good results and allow data to be successfully extracted. However, low-quality images, images with noise (blurred, low/high-lighting, etc), images that are rotated, or images taken at a long-distance, can severely degrade the IRIS's performance.
To achieve better results when using IRIS, we recommend requesting high-quality images from the end-users. There may still be cases where IRIS can fail on text extraction from high-quality images, due to the text font used, text orientation, tiny text, and so on. There may also be some cases of characters being confused during text extraction. For instance, instead of recognizing the number "1" it recognizes the letter "l" and, consequently, results in a failure during the field classification phase.
Therefore, you should consider including a confirmation step in your processes to guarantee that all information is accurate.
To learn more or to help you get started with IRIS, feel free to contact Sinch or if you are an existing customer reach out to your Sinch Account Manger.