Essam, Yes. They are all real images. We're using web scraping to collect the images from Google, Bing, Pinterest, Instagram... We're using LSTM with an Attention layer to make sure OCR will work even if the MICR lines are mixed with the signature, stamps, annotations... There is an online webapp to check the accuracy at https://www.doubango.org/webapps/micr/
On Saturday, April 4, 2020 at 11:59:34 AM UTC+2, Essam Zaky wrote: > > Hi @mamadou > > how did you collected the 17000 image are they real images , > also which type of Tensorfolw models you used , LSTM line , or single > character model > > Best Regards > Essam > > بتاريخ الخميس، 2 أبريل، 2020 8:22:44 م UTC+2، كتب Ghada Aruri: >> >> Hi team, >> >> For CMC-7, I want to train it by using jTessBoxEditor to get >> cmc7.traineddata what the steps to get the cmc7.traineddata? >> and if anybody has done it and is willing to share me if you can? >> >> Best Regards. >> > -- You received this message because you are subscribed to the Google Groups "tesseract-ocr" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/a73ee04c-358b-469f-9b77-5bf71f8ade67%40googlegroups.com.
