Open KonstantinDjairo opened 10 months ago
90% of the time it gets the wrong result, it's almost impossible to use. There's a project in which they used a manga dataset to train the model and it worked great.
https://github.com/kha-white/manga-ocr
i wonder if kaku could use that model or use the dataset to train a specifc model to be used in the app.
90% of the time it gets the wrong result, it's almost impossible to use. There's a project in which they used a manga dataset to train the model and it worked great.
i wonder if kaku could use that model or use the dataset to train a specifc model to be used in the app.