I'm trying to use Optical Character Recognition to extract text from images of cattle tags like . The tags are yellow with black text, and they may be slightly distorted or rotated. I'm wondering what the best way to pre-process these images for EasyOCR (or another OCR tool) would be. Here are some specific questions I have: What are some good pre-processing techniques for improving the accuracy of OCR on these types of images? Which OCR engine would be best suited for this task? Would it be beneficial to train a custom OCR model on a dataset of cattle tag images? I've done some initial research on image pre-processing for OCR, but I'm not sure which techniques would be most effective for this specific case. Any advice would be greatly appreciated. I tried using EasyOCR and got 1 or 2 images correctly out of 18 but that is not enough. Continue reading...
