Share This Author
Deepfake Video Detection Using Recurrent Neural Networks
A temporal-aware pipeline to automatically detect deepfake videos is proposed that uses a convolutional neural network to extract frame-level features and a recurrent neural network that learns to classify if a video has been subject to manipulation or not.
Perceptual watermarks for digital images and video
Digital image watermarking techniques known as perceptually watermarks that are designed to exploit aspects of the human visual system in order to produce a transparent, yet robust watermark are described.
First Steps Toward Camera Model Identification With Convolutional Neural Networks
- L. Bondi, L. Baroffio, David Guera, Paolo Bestagini, E. Delp, S. Tubaro
- Computer ScienceIEEE Signal Processing Letters
- 3 March 2016
A data-driven algorithm based on convolutional neural networks, which learns features characterizing each camera model directly from the acquired pictures, which outperforms up-to-date state-of-the-art algorithms on classification of 64 × 64 color image patches.
Digital watermarking: algorithms and applications
A general framework for watermark embedding and detection/decoding is presented here along with a review of some of the algorithms for different media types described in the literature.
A Two Stream Siamese Convolutional Neural Network for Person Re-identification
- Dahjung Chung, Khalid Tahboub, E. Delp
- Computer ScienceIEEE International Conference on Computer Vision…
- 1 October 2017
This paper presents a two stream convolutional neural network where each stream is a Siamese network that can learn spatial and temporal information separately and proposes a weighted two stream training objective function which combines theSiamese cost of the spatial and spatial streams with the objective of predicting a person's identity.
Printer identification based on graylevel co-occurrence features for security and forensic applications
- A. Mikkilineni, Pei-Ju Chiang, G. N. Ali, G. Chiu, J. Allebach, E. Delp
- Computer ScienceIS&T/SPIE Electronic Imaging
- 21 March 2005
The use of image texture analysis to identify the printer used to print a document is described and a set of features that can be used to provide forensic information about a document are described.
The EM/MPM algorithm for segmentation of textured images: analysis and further experimental results
New theoretical results show that the EM/MPM algorithm can be expected to achieve the goal to minimize the expected value of the number of misclassified pixels, to the extent that theEM estimates of the model parameters are close to the true values of themodel parameters.
Error concealment in MPEG video streams over ATM networks
- P. Salama, N. Shroff, E. Delp
- Computer ScienceIEEE Journal on Selected Areas in Communications
- 1 June 2000
A technique for packing ATM cells with compressed data is described, whereby the location of missing macroblocks in the encoded video stream can be found and permits the proper decoding of correctly received macroblocks, and thus prevents the loss of ATM cells from affecting the decoding process.
A watermark for digital images
- Raymond B. Wolfgang, E. Delp
- Computer ScienceProceedings of 3rd IEEE International Conference…
- 16 September 1996
This work describes two techniques for the invisible marking of images and analyzes the robustness of the watermarks with respect to linear and nonlinear filtering, and JPEG compression.
Multiresolution detection of spiculated lesions in digital mammograms
Experimental results using the MIAS image database have shown that the proposed multiresolution scheme is capable of detecting spiculated lesions of very different sizes at low false positive rates.