I have dual stream scans for 4 fixed form documents types , each copy is scanned at the same resolution and dimensions - 1 color copy / 1 red color drop out b/w copy,
I train the fixed form templates based on the color (bw version) because all of the separators are there and switch to the color drop out before recognition.
two of the documents are nearly similar and the differentiating features are only present in the color copy,
what are some of the tricks you've used to enhance recognition for the two types,
more so what are some of the tricks you've used to figure out what the classification step was doing ?
I train the fixed form templates based on the color (bw version) because all of the separators are there and switch to the color drop out before recognition.
two of the documents are nearly similar and the differentiating features are only present in the color copy,
what are some of the tricks you've used to enhance recognition for the two types,
more so what are some of the tricks you've used to figure out what the classification step was doing ?
Comments
0 comments
Please sign in to leave a comment.