Home > Events > PhD Defense - Aparna Bharati

PhD Defense - Aparna Bharati

Start: 7/2/2020 at 10:30AM
End: 7/2/2020 at 1:30PM
Location: Remote via Zoom
Attendees: TBA
Add to calendar:
iCal vCal

Aparna Bharati

Dissertation Defense

July 2, 2020        10:30 am        Remote via Zoom

Adviser:  Dr. Kevin Bowyer

Committee Members:

    Dr. Patrick Flynn      Dr. Nasir Memon      Dr. Anderson Rocha

Dr. Walter Scheirer

Title:

"Vision and Learning-Based Methods for Scalable and Generalized Image Forensics"

 Abstract:

Image Forensics has become an important area of research due to the exponential increase in availability and free exchange of media. The accessibility to good quality sensors and image hosting websites is now at the disposal of more users than ever. The increase in exchange of images has also led to more ways to edit image content aimed towards achieving a certain goal. Whether it is altering one's portrait to look younger, increasing color contrast in natural scenes or adding external objects to the images to change the perception and understanding of images, improved and freely available image editing software can allow a non-technical person to create images with changes undetectable to the naked eye. Edits or manipulations with malicious intent can be used to mislead readers or followers. Certain kinds of usage create social and legal concerns with respect to the spread of misinformation through the media while others can skew people's perception of reality. Images manipulated with benign intent behind them may also be fatal to the society by shifting the normal of visual information and the expectations of viewers.
 
In order to assess and regulate the quality of media, it is important to devise algorithms that detect and analyze manipulated content in an automated way. This body of work focuses on solving two such problems - detecting retouching effects in face images and analyzing the provenance of a given image. The former one detects appearance-based manipulation in face images of individuals while the latter implies understanding the evolution history of an image. In addition, to understanding single image properties, the latter analysis considers a collection of active variants of the image in question and requires retrieving the stages of evolution of the manipulated media object and the other objects contributing parts to the stages. The proposed methods used to solve the problems focus on efficiency and generalizability as the scale of operation is quite large and the problem is very unconstrained. This work contributes towards formalizing the problem definition and creating solutions that are applicable to general cases of manipulated images and on a large scale. The methods proposed to improve multi-demographic operation for retouching detection in faces. Algorithms developed for purposes of Image Provenance Analysis are flexible for images in terms of content, quality, and source. They are evaluated on unconstrained scenarios and tested with large scale datasets. The pipelines proposed for solving these problems from the domain of image forensics utilize techniques from image analysis, pattern recognition, and computer vision.