CS297 Proposal
Neural net Captcha Cracker.
Geetika Garg(geetikagarg07@gmail.com)
Advisor: Dr. Chris Pollett
Description:
A CAPTCHA (an acronym for "Completely Automated Public Turing test to tell Computers and Humans Apart") is a type of challenge-response test used in computing to determine whether or not the user is human. In this project, I will be working with Dr. Pollett on neural networks to crack CAPTCHAs.
In the end, we will be able to make a program removes the effects and noise from an image and then will use neural nets to recognize a CAPTCHA. We will be training a large sample of data on neural nets to build our OCR.
Schedule:
Week 1: Jan. 22 - 28 | Meet Professor and set the schedule for future meetings |
Week 2:
Jan 29- Feb 4 | Get the site up |
Week 3:
Feb 5 - Feb 11 | Will read about the image processing |
Week 4:
Feb 12 - Feb 18 | Will write a python script to read an image and clean it. |
Week 5:
Feb 19 - Feb 25 | Deliverable 1: Python program to read and clean up image. |
Week 6:
Feb 26-Mar 4 | Will read about the literature behind CAPTCHAs |
Week 7:
Mar 5 - Mar 11 | Deliverable 2: Literature review on CAPTCHAs |
Week 8:
Mar 12 - Mar 18 | Learn the basics of OCR |
Week 9:
Mar 19 - Mar 25 | Continue the work from week 9 |
Week 10:
Mar 26 - Apr 1 | Deliverable 3: Program implementing OCR |
Week 11:
Apr 2 - Apr 8 | Learn to train sample dataset using neural nets |
Week 12:
Apr 9 - Apr 15 | Continue the work from week 11 |
Week Apr 16 - Apr 22 | Deliverable 4:Will show the results of training the data. |
Week 14:
Apr 23 - Apr 29 | Will start working on CS297 report |
Week 15:
Apr 30 - May 6 | Continue with work from week 14 |
Week 16:
May 7 - May 13 | Deliverable 5: CS297 report |
Deliverables:
The full project will be done when CS298 is completed. The following will
be done by the end of CS297:
1. Python program to read and clean up image
2. Literary review on Captchas
3. Program implementing OCR
4. Results of training a data set through neural nets.
5. Complete 297 report.
References:
David R. Martin, Charless C. Fowlkes, Jitendra Malik(2002).Learning to Detect Natural Image Boundaries
Using Brightness and Texture.U.C. Berkeley, Berkeley, CA 94720
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.50.6383&rep=rep1&type=pdf |