Confidence levels

During recognition, Datacap assigns a confidence level to each character and OMR field. Confidence levels range from 1 (lowest confidence) to 10 (highest confidence).

You can see the confidence level for each character or OMR field in the cn attribute of the object in the page data file.

<F id="Pickup_Date">
   <V n="TYPE">Pickup_Date</V>
   <V n="Position">189,403,567,465</V>
   <V n="STATUS">0</V>
   <C cn="7" cr="205,414,219,439">83</C>				<-- ASCII 'T' [low confidence]
   <C cn="4" cr="205,414,219,439">83</C>				<-- ASCII 'r' [low confidence]
   <C cn="10" cr="224,423,236,438">117</C>			<-- ASCII 'u' [high confidence]
   <C cn="10" cr="241,423,255,438">101</C>			<-- ASCII 'e' [high confidence]
   <C cn="10" cr="256,423,266,438">115</C>			<-- ASCII 's' [high confidence]
   <C cn="10" cr="270,434,275,441">44</C>				<-- ASCII ',' [high confidence]
   <C cn="10" cr="334,416,335,438">32</C>				<-- ASCII ' ' [high confidence]
   <C cn="10" cr="288,416,304,438">68</C>				<-- ASCII 'D' [high confidence]
   <C cn="10" cr="308,423,320,438">101</C>			<-- ASCII 'e' [high confidence]
   etc.
</F>

The confidence level determines how Datacap displays the character and the parent field within the verification panel:

  • The two verification clients display fields that contain low confidence characters in yellow, where low confidence in this case is anything less than 10.
  • Within the field, the Datacap Web Client displays the low confidence characters in red, while Datacap Desktop highlights the problem characters in yellow within the image snippet. Low confidence in this case is anything less than 10 or the field's ReqConf value. (See Overriding the default confidence value on specific fields.)