learn existing OCR implementation

[ttaliacc]

Before we implement try to implement and embed the ML model to the quick entry form, we should firstly find out how the OCR works. It will be ideal for us to synchronize our implementation with OCR's.

[ttaliacc]

checkout this doc

https://biokic.github.io/symbiota-docs/editor/edit/ocr/

[ttaliacc]

The OCR engines are Tesseract and Digi-Leap

related files

• collections/editor/includes/imgprocessor.php: use ocrImage to trigger the function
• js/symb/collections.editor.imgtools.js: use Sends this data (image ID, region, mode, etc.) to the server via AJAX to process the image. After the AJAX request completes, it receives the OCR text, updates the relevant UI elements, and re-enables the button.
• collections/editor/rpc/ocrimage.php: backend component responsible for handling OCR requests from the frontend. When an image is sent for OCR processing, this script receives the request, processes the image using Tesseract or Digi-Leap accordingly. It also allows to get crop of images for different OCR use cases.
• classes/SpecProcessorOcr.php: This class manages the OCR process, including loading the image, cropping, and interacting with Tesseract and Digi-Leap.

[ttaliacc]

We found the OCR is actually a local running instance. To run it, the user will need to install the Tesseract locally, and the funcion private function ocrImageViaTesseract($url = "") of classes/SpecProcessorOcr.php will pass the URL to the instance and use the OCR. This is also why we were not able to run the OCR of the old entry form but also didn't see any error in log