At the entrance - scanned files of documents of different quality of any subject. The quality of input documents is different, scans 600dpi.
The output is a document in Excel, which is very similar to a scan.
All information fields (all tables and text) must be correctly recognized.
Basic recognition of information with an accuracy of at least 97%.
The system must be trained and improve recognition accuracy.
Code in C ++ or C #.
Software should work autonomously. Work with clouds is not allowed.
18 pekerja bebas membida secara purata $3142 untuk pekerjaan ini
I am a co-founder of an Artificial intelligence software startup that works on Image recognition,Machine learning,face recognition and speech recognition systems. Will code this in C++.