What is OCR?
OCR is a technology that allows you to access the text or writing that appears in an image. With AppSheet's OCR feature, you can speed up data entry by automatically filling out your forms from an image.
Watch a video to learn more about OCR.
What kinds of images work with OCR?
AppSheet's OCR feature can be used when you need to scan multiple documents that all have a fixed layout. The image may contain printed text or handwriting. Some examples of this include:
- A standardized paper form
- USPS shipping labels
- Invoices from a single company
Some cases where AppSheet's OCR feature will not work include:
- Business cards (which do not adhere to a standard format)
- Forms that use checkboxes
- Apps that do not have an internet connection
How do I use OCR in my app?
AppSheet's OCR feature is currently in beta. You can access it in the AppSheet editor under Intelligence > OCR.
To get started with OCR, you will need at least four example images and the values that should be extracted from them. AppSheet will inspect the given examples and come up with a general template to match against new images.
Once you have your examples, you can create a new OCR model:
- Open your app in the editor.
- Go to the Intelligence > OCR Models pane.
- Click + New OCR Model.
- Select the table that contains the image examples in the Table drop-down.
- Choose the image column that has the examples inputs in the Image Column drop-down.
- Set the Output Columns to include the columns that contain the extracted data.
Save the app by selecting one of the following options:
- Save - Save the app
- Save & verify data - Save the app and verify the data defined.
- Save & notify users - Save the app and notify users.
Once you save your changes, your custom OCR model will be created. AppSheet will show the patterns that were found in each example and inform you of any issues that happened along the way.
For example, let's say we're extracting information from pictures of a conference badge. We'll create a new OCR model that will operate on the
Badges table, extracting from the
Badge image column, and returning the
Twitter Handle, and
Role from the image.
After the OCR model has finished training, it's available for use in your app. Your columns will automatically be initialized using the extracted values from the image. Behind the scenes, AppSheet has added special OCR formulas to initialize the values of your columns.