We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: MailLeak: Obfuscation-Robust Character Extraction Using Transfer Learning

Abstract: The following work presents a new algorithm for character recognition from obfuscated images. The presented method is an example of a potential threat to current postal services. This paper both analyzes the efficiency of the given algorithm and suggests countermeasures to prevent such threats from occurring.
Subjects: Machine Learning (cs.LG); Cryptography and Security (cs.CR)
Cite as: arXiv:2012.11775 [cs.LG]
  (or arXiv:2012.11775v1 [cs.LG] for this version)

Submission history

From: Hugues Nelson Iradukunda [view email]
[v1] Tue, 22 Dec 2020 01:14:28 GMT (650kb)

Link back to: arXiv, form interface, contact.