You may have noticed that there are several different recognition modes available to choose from for Long Text and Short Text question types. But which one should you choose? Well, it depends on your use case.

If you only want to recognize numbers, please skip to the Digit recognition section.

Text Recognition (Long Text and Short Text)

Full: Recognize text & flag to verify

This is the default mode for the text recognition and it is likely the one you would want to use.

• It runs the captured fields through handwriting text recognition software.
• All responses will be flagged for manual verification allowing you to review and correct them if needed.
• Handwritten text is easier to recognize when it is WRITTEN IN CAPITAL LETTERS.

Why you might not want to use it

• Handwritten text recognition isn't 100% accurate. Therefore, you will need to review the responses.
• It is best for reading sentences, it will not detect individual letters or numbers.
• Your account has a limit of how many handwritten fields you can read each month as we are relying on external service for handwritten text verification.

Partial: Recognize text & skip verification

• It will run the fields through the handwriting text recognition software.
• It will not flag the responses for the additional verification.
• This is suitable if you are only interested in quantitative data but you do not care about digitizing text responses accurately as you can always read the responses from the image.

Manual: Do not recognize & verify manually

• The text fields will not run through handwriting recognition.
• All fields will be flagged for you to manually verify
• This setting will be automatically applied if you go above the limit of fields you can recognize per month.

None: Do not recognize text & skip verification

• The text fields will not run through handwriting recognition.
• Responses will not be flagged for verification
• Suitable if you are only interested in quantitative data but you do not care about digitizing text responses as you can always read the text responses from the image.

OCR: Read the machine-printed text

Instead of using handwritten text recognition you can use OCR (Optical character recognition) technology to recognize the printed answers.

• It will only read machine-printed characters.
• Handwritten text is not going to be recognized.
• This method is useful if you would like to pre-fill the field before printing, use date stamps, info-daters or stickers to add machine-printed text to printed forms.

ICR: Read handwritten characters

**The ICR (Intelligent Character Recognition) technology is currently available, but it is in the BETA stage. It may not recognize some of the characters, but we are working on improving the edge cases.

Digit recognition (Number and Date)

Date and Number question types use an ICR (Intelligent Character Recognition) technology to read digits. It is a very different technology from the (HTR) handwritten text recognition as it only needs to recognize 10 different digits. Additionally, it is much more accurate. If digits are accurately predicted, they will not be flagged for additional verification.

• Digit recognition does not count towards your handwriting field recognition limit.
• Much more accurate than handwriting recognition.
• Most of fields are recognized correctly.

While you may select to display the field as a free-text response, we highly recommend that you specify the maximum number of digits the respondent is allowed to enter. If not all the boxes are filled, this is okay - the empty boxes will be automatically ignored.

Specifying the number of digits makes the detection much more accurate. Having a free-text response area may make it difficult to detect numbers written very close to each other. If you still want to keep the free-text box, make sure you leave plenty of space between the digits.