Hi @alirida.alhalass ,
Could you try applying OCR on these images and Check what is the result that you will be receiving ?
More so, You could directly adopt the Document Understanding Framework and modify parts of the workflow to adjust to your requirements.
You would require to perform Labelling of these images in the Data Labelling session. Use the Document Understanding ML Package and train the model based on these input data.
Deploy the Model and check the accuracy of the extraction.
Although the image quality looks pretty good, you might need to train with more samples for the model to correctly recognise the value to be extracted.