MediFor (Media Forensics)
The Media Forensics research project (MediFor) is an endeavor funded by the Defense Advanced Research Projects Agency (DARPA) and the Air Force Research Laboratory (AFRL) whose goal is to develop solutions for the automated assessment of the integrity of digital images. More details can be obtained here: https://www.darpa.mil/program/media-forensics.
Working together with the Universities of Purdue, South California (USC), New York (NYU), Siena, Campinas (Unicamp), and the Politécnico di Milano, the CVRL team leads the development of solutions to the problem of Provenance Analysis. Given a questioned image, namely a probe, and a large corpus of images (such as the Internet), Provenance Analysis aims at two major tasks:
(1) Finding the images that directly and transitively share content with the probe (a task we call Provenance Image Filtering).
(2) Building the directed acyclic graph whose nodes individually represent the probe and related images, and whose edges express the edition and content-donation history (e.g., cropping, blurring, removal, splicing, etc.) between pairs of images, linking seminal to generated elements (a task we call Provenance Graph Construction)
By combining ideas from the areas of image retrieval, digital image forensics, and graph theory, Provenance Analysis constitutes an interesting interdisciplinary topic that spans the fields of image processing and computer vision. Watch the video above for more details.
Aparna Bharati, Joel Brogan, Daniel Moreira, Kevin Bowyer, Patrick Flynn, Anderson Rocha, Walter Scheirer
News: [McClatchy] [NBC News] [GCN] [Purdue University]
Image Provenance Analysis at Scale, Daniel Moreira, Aparna Bharati, Joel Brogan, Allan Pinto, Michael Parowski, Kevin W Bowyer, Patrick J Flynn, Anderson Rocha, Walter J Scheirer, 2018: [arxiv] [pdf] [video]
Provenance Filtering for Multimedia Phylogeny, Allan Pinto, Daniel Moreira, Aparna Bharati, Joel Brogan, Kevin Bowyer, Patrick Flynn, Walter J. Scheirer, Anderson Rocha, Proceedings of the IEEE International Conference on Image Processing (ICIP), September 2017: [pdf] [code]
U-Phylogeny: Undirected Provenance Graph Construction in the Wild, Aparna Bharati, Daniel Moreira, Allan Pinto, Joel Brogan, Kevin Bowyer, Patrick Flynn, Walter J. Scheirer, Anderson Rocha, Proceedings of the IEEE International Conference on Image Processing (ICIP), September 2017: [pdf] [code] [poster]
Spotting the Difference: Context Retrieval and Analysis for Improved Forgery Detection and Localization, Joel Brogan, Paolo Bestagini, Aparna Bharati, Allan Pinto, Daniel Moreira, Kevin Bowyer, Patrick Flynn, Anderson Rocha, Walter J. Scheirer, Proceedings of the IEEE International Conference on Image Processing (ICIP), September 2017: [pdf] [code] [poster]
Beyond Pixels: Image Provenance Analysis Leveraging Metadata, Aparna Bharati, Daniel Moreira, Joel Brogan, Patricia Hale, Kevin Bowyer, Patrick Flynn, Anderson Rocha, Walter J. Scheirer, IEEE Winter Conference on Applications of Computer Vision (WACV), January 2019 [arxiv]
Dynamic Spatial Verification for Large-Scale Object-Level Image Retrieval, Joel Brogan, Aparna Bharati, Daniel Moreira, Kevin Bowyer, Patrick Flynn, Anderson Rocha, Walter Scheirer, December 2019: [arxiv]
Learning Transformation-Aware Embeddings for Image Forensics, Aparna Bharati, Daniel Moreira, Patrick Flynnm, Anderson Rocha, Kevin Bowyer, Walter Scheirer, January 2020: [arxiv]
