Can OCR works on this form?

Im planning to have a dummy table with the following columns in the sample image below. The task is to upload this image (positions remain static) and the OCR will identify each column and populate my dummy table as a temporary storage accordingly. User will later assign each row a name and commit to the actual table.

 

Would this be possible with OCR function in Appsheet? Any idea how to do this?

 

Thanks in advance!

1000023014.jpg

Solved Solved
0 4 115
1 ACCEPTED SOLUTION

At the outset, let me reiterate that OCR feature is in beta and has several caveats as mentioned in the article at the moment.

Optical Character Recognition (OCR) - AppSheet Help

Having said that you could give it a try by training it on minimum of 4 rows of data. Again I believe the input images will need to have a consistency , clarity and firm positioning of the input data. meaning if there is a name field , it will need to be at the same position in all input data being fed as image. 

Please note : you may need to experiment at your judgement/time and it may or may not be useful for your requirement.

View solution in original post

4 REPLIES 4

Btw, my dummy table has always 12 rows with dummy names and default 0 values in each columns every after.commit or initial load up of the form.

It is not clear if the dummy table is a spreadsheet or in the app itself. 

If the requirement is that the dummy table is in a spreadsheet and you would like it to be uploaded to a table in AppSheet, one option you could try is CSV import.

Import a CSV file for this view - AppSheet Help

OCR is currently under beta and may not be very suitable for a multirow read as per my understanding. Its typical good use case is for reading a row level information such as say a visiting card or a name badge.

Suvrutt thanks! My dummy table will be in google sheet that has 12 rows with empty values. Im wondering if OCR could populate these rows accordingly. If OCR can capture one row at a time, can we train it?,

At the outset, let me reiterate that OCR feature is in beta and has several caveats as mentioned in the article at the moment.

Optical Character Recognition (OCR) - AppSheet Help

Having said that you could give it a try by training it on minimum of 4 rows of data. Again I believe the input images will need to have a consistency , clarity and firm positioning of the input data. meaning if there is a name field , it will need to be at the same position in all input data being fed as image. 

Please note : you may need to experiment at your judgement/time and it may or may not be useful for your requirement.

Top Labels in this Space