writer dependent offline recognition system
Note Description
This research paper approaches the problem from a feature learning angle. The hypothesis is that it is smarter to take in the features from data, rather than utilizing hand-made features that have no similarity to the signature generation process. To prove this, Deep Convolutional Neural Network was used to learn the features in a writer-independent format; then the model is used to get a feature representation on a different set of users, where the writer-dependent classifiers are trained. The GPDS-960 and CEDAR datasets were used.
...
Notes information
| Seller Price | GHC 5 |
| Added | 26 Oct, 2020 |
| University | Ashesi University |
| Course | Computer Science |
Note Review
Zylac Webek
3 years ago
Worked on a similar project in school, by far the best research I have seen on offline recognition system