That ResNet-example from ML reminded me that I could do something similar for the texts of one of my grandmas. I love receiving letters and postcards from her, but I have problems with deciphering :')

So why not take that challenge and create a tool, which could convert the photo of a text block into readable text?

• grayscale and preprocess the image (maybe reduce resolution)
• find and extract all image parts which contain handwritten letters (as block or one after another?)
• pre-train a model (ML) to recognize all samples (prepare dataset for this)

• good, but has strong issues: https://www.pyimagesearch.com/2020/08/24/ocr-handwriting-recognition-with-opencv-keras-and-tensorflow/
• extract handwriting: https://stackoverflow.com/questions/58141153/how-do-i-isolate-handwritten-text-from-an-image-using-opencv-and-python
• http://cs231n.stanford.edu/reports/2017/pdfs/810.pdf

• promising approach with ResNet101: https://medium.com/@akshaychougule/handwritten-devanagari-character-identification-using-resnet-b90894b42c4d

• almost forgotten about this: https://modelzoo.co/
• ugh: https://nanonets.com/blog/handwritten-character-recognition/
• google vision as python-lib can be helpful to extract the text: https://stackoverflow.com/questions/58805719/how-to-extract-both-automated-and-handwritten-text-in-image-using-gcp-vision-or *how to find a baseline of written text: https://www.researchgate.net/post/How_to_find_the_baseline_of_a_handwritten_word