SubhrajyotiSen/KannadaHandwritingRecognition ? reverse-engineered prompt

Build me a web app that can take a handwritten Kannada document image and turn it into digital text.

I want the full flow working, from uploading an image, preprocessing it, segmenting it into lines, words, and characters, then recognizing the text and showing the result in the browser. It should handle Kannada handwriting well, including tricky character combinations and ottaksharas, and it should also do some basic cleanup like grayscale, noise removal, contrast normalization, binarization, and character isolation.

If it makes sense, include multiple recognition options like k nearest neighbors, SVM, CNN, and Inception style model so I can compare them. Please make the app easy to run locally, and if you need any current library details or model setup info, look them up online. A simple upload page with a clear output area would be great, along with whatever training or inference code is needed to make it all work end to end.