Optical character recognition (OCR) extracts text presented in an image. However, it doesn’t provide any additional formatting capabilities. If you want to extract information in JSON format, you can only do so if the data presented is already in JSON. This typically isn’t the case, but there’s an easy way to go from OCR to JSON in just a few steps — AI.
Converting extracted information into JSON usually requires coding. However, generative AI can intelligently do so without requiring users to know any syntax or markup. Instabase uses generative AI and large language models so that anyone can simply use natural language to extract and convert data from documents and images, among other functionalities. With Instabase’s Converse app, users can extract and convert data from documents one by one, while its Build app is made for users who need to execute this task frequently or have large volumes of data to process.
How to Use Instabase Converse for OCR to JSON
Instabase Converse lets you upload documents and easily extract information from them in various formats, including JSON — just follow these directions.
- Go to aihub.instabase.com and select the option to “Create a chat” to open Converse.
- Click on “Add files” in the bottom-left corner to upload the document that you’d like to extract and convert data from, and either log in to your existing Instabase account or create a new one. For this example, we’ll be using the sample resume that’s already uploaded in Converse.
- In the text box located in the bottom right, tell Converse to extract the information you want and format it in JSON.
4. You can copy and paste the results by hovering over the top-right corner of the output and clicking the two overlapping squares. Or, click the arrow to download the output in a TXT file.
How to Use Instabase Build to Automate the OCR to JSON Workflow
If you need to convert extracted data into JSON on a regular basis or you’re working with multiple documents, use the Build app to create an automated, repeatable workflow for it. Here’s how to build your own OCR to JSON app — no coding required.
- Go to aihub.instabase.com and select the “Create an app” option. You’ll be prompted to either create an Instabase account if you’re a new user or log in to your Instabase account if you’re a current user.
- Upload one to five documents that you’d like to extract data from.
- Click the “Create classes” icon in the upper right.
4. Enter a class name that reflects the type of document you’re working with, or select from a suggested class name. In this example, we’re using “Resume” as the class name. You can also add an optional description.
5. Hit “Enter” on your keyboard and then click the “Classify documents” in the bottom-right corner.
6. You may need to click “x” in the upper-right corner of the right panel to see the “Add field” button. Click “Add field.”
7. Enter a field name for a data field that you’d like to extract, or select from one of the suggestions. You’ll create additional field names one at a time.
8. Change the field type to “Document reasoning.”
9. Enter a natural language prompt that tells Build to extract the field and format it in JSON. Then click “Run” to update the result.
10. Confirm that the new result is correct and click “x” in the upper-right corner to save the field. If it’s not correct, try changing your prompt or selecting a different model.
11. Hovering over each created field gives you the option to copy the extracted data, edit the field, reorder the field, or delete the field. When you’re done adding fields, click “Create app” in the upper-right corner.
12. Give your app a name and an optional description, icon, and sample files. Click “Next.”
13. Choose your release state and add optional release notes. Click “Create app.”
14. Click “Open app.”
15. Click “Run app.”
16. Upload the files that you’d like to apply your new app to.
17. Once your files have been successfully uploaded, click “Run.”
18. When the run has been completed, click the name of the run to view the results.
19. You should now see the extracted data in the right panel. Hovering your mouse over any of the extracted data brings up the option to copy the data. You can also click “Export results” in the upper-right corner to download the data or save it to an external drive.
OCR to JSON in Seconds
Automate data extraction and conversion with Instabase’s AI applications and do more in less time.
