With advent of newer architectures in Deep Learning, accuracies achieved have been close to 100% in the domain of Optical Character Recognition of Numbers. But computational complexity is a challenge that needs to be addressed. Can anyone suggest the most accurate and lightest model that can be incorporated in any web app and give instant results of OCR?

More Avishek Garain's questions See All
Similar questions and discussions