U. Dinesh Acharya

Jiqing Wu, Zhiwu Huang, Dinesh Acharya et al.

In generative modeling, the Wasserstein distance (WD) has emerged as a useful metric to measure the discrepancy between generated and real data distributions. Unfortunately, it is challenging to approximate the WD of high-dimensional distributions. In contrast, the sliced Wasserstein distance (SWD) factorizes high-dimensional distributions into their multiple one-dimensional marginal distributions and is thus easier to approximate. In this paper, we introduce novel approximations of the primal and dual SWD. Instead of using a large number of random projections, as it is done by conventional SWD approximation methods, we propose to approximate SWDs with a small number of parameterized orthogonal projections in an end-to-end deep learning fashion. As concrete applications of our SWD approximations, we design two types of differentiable SWD blocks to equip modern generative frameworks---Auto-Encoders (AE) and Generative Adversarial Networks (GAN). In the experiments, we not only show the superiority of the proposed generative models on standard image synthesis benchmarks, but also demonstrate the state-of-the-art performance on challenging high resolution image and video generation in an unsupervised manner.

Dinesh Acharya, Zhiwu Huang, Danda Pani Paudel et al.

The extension of image generation to video generation turns out to be a very difficult task, since the temporal dimension of videos introduces an extra challenge during the generation process. Besides, due to the limitation of memory and training stability, the generation becomes increasingly challenging with the increase of the resolution/duration of videos. In this work, we exploit the idea of progressive growing of Generative Adversarial Networks (GANs) for higher resolution video generation. In particular, we begin to produce video samples of low-resolution and short-duration, and then progressively increase both resolution and duration alone (or jointly) by adding new spatiotemporal convolutional layers to the current networks. Starting from the learning on a very raw-level spatial appearance and temporal movement of the video distribution, the proposed progressive method learns spatiotemporal information incrementally to generate higher resolution videos. Furthermore, we introduce a sliced version of Wasserstein GAN (SWGAN) loss to improve the distribution learning on the video data of high-dimension and mixed-spatiotemporal distribution. SWGAN loss replaces the distance between joint distributions by that of one-dimensional marginal distributions, making the loss easier to compute. We evaluate the proposed model on our collected face video dataset of 10,900 videos to generate photorealistic face videos of 256x256x32 resolution. In addition, our model also reaches a record inception score of 14.57 in unsupervised action recognition dataset UCF-101.

Dinesh Acharya, Zhiwu Huang, Danda Paudel et al.

Classifying facial expressions into different categories requires capturing regional distortions of facial landmarks. We believe that second-order statistics such as covariance is better able to capture such distortions in regional facial fea- tures. In this work, we explore the benefits of using a man- ifold network structure for covariance pooling to improve facial expression recognition. In particular, we first employ such kind of manifold networks in conjunction with tradi- tional convolutional networks for spatial pooling within in- dividual image feature maps in an end-to-end deep learning manner. By doing so, we are able to achieve a recognition accuracy of 58.14% on the validation set of Static Facial Expressions in the Wild (SFEW 2.0) and 87.0% on the vali- dation set of Real-World Affective Faces (RAF) Database. Both of these results are the best results we are aware of. Besides, we leverage covariance pooling to capture the tem- poral evolution of per-frame features for video-based facial expression recognition. Our reported results demonstrate the advantage of pooling image-set features temporally by stacking the designed manifold network of covariance pool-ing on top of convolutional network layers.

Jiqing Wu, Zhiwu Huang, Janine Thoma et al.

In many domains of computer vision, generative adversarial networks (GANs) have achieved great success, among which the family of Wasserstein GANs (WGANs) is considered to be state-of-the-art due to the theoretical contributions and competitive qualitative performance. However, it is very challenging to approximate the $k$-Lipschitz constraint required by the Wasserstein-1 metric~(W-met). In this paper, we propose a novel Wasserstein divergence~(W-div), which is a relaxed version of W-met and does not require the $k$-Lipschitz constraint. As a concrete application, we introduce a Wasserstein divergence objective for GANs~(WGAN-div), which can faithfully approximate W-div through optimization. Under various settings, including progressive growing training, we demonstrate the stability of the proposed WGAN-div owing to its theoretical and practical advantages over WGANs. Also, we study the quantitative and visual performance of WGAN-div on standard image synthesis benchmarks of computer vision, showing the superior performance of WGAN-div compared to the state-of-the-art methods.

Jiqing Wu, Zhiwu Huang, Dinesh Acharya et al.

In generative modeling, the Wasserstein distance (WD) has emerged as a useful metric to measure the discrepancy between generated and real data distributions. Unfortunately, it is challenging to approximate the WD of high-dimensional distributions. In contrast, the sliced Wasserstein distance (SWD) factorizes high-dimensional distributions into their multiple one-dimensional marginal distributions and is thus easier to approximate. In this paper, we introduce novel approximations of the primal and dual SWD. Instead of using a large number of random projections, as it is done by conventional SWD approximation methods, we propose to approximate SWDs with a small number of parameterized orthogonal projections in an end-to-end deep learning fashion. As concrete applications of our SWD approximations, we design two types of differentiable SWD blocks to equip modern generative frameworks---Auto-Encoders (AE) and Generative Adversarial Networks (GAN). In the experiments, we not only show the superiority of the proposed generative models on standard image synthesis benchmarks, but also demonstrate the state-of-the-art performance on challenging high resolution image and video generation in an unsupervised manner.

Suma D., U. Dinesh Acharya, Geetha M. et al.

Locating and distilling the valuable relevant information continued to be the major challenges of Information Retrieval (IR) Systems owing to the explosive growth of online web information. These challenges can be considered the XML Information Retrieval challenges as XML has become a de facto standard over the Web. The research on XML IR starts with the classical IR strategies customized to XML IR. Later novel IR strategies specific to XML IR are evolved. Meanwhile literatures reveal development of the rapid and intelligent IR systems. Despite their success in their specified constrained domains, they have additional limitations in the complex information space. The effectiveness of IR systems is thus unsolved in satisfying the most. This article attemptsan overview of earlier efforts and the gaps in XML IR.

S. Hemalatha, U. Dinesh Acharya, A. Renuka

Steganography is one of the methods used for secret communication. Steganography attempts to hide the existence of the information. The object used to hide the secret information is called as cover object. Images are the most popular cover objects used for steganography. Different techniques have to be used for color image steganography and grey scale image steganography since they are stored in different ways. Color image are normally stored with 24 bit depth and grey scale images are stored with 8 bit depth. Color images can hold large amount of secret information since they have three color components. Different color spaces namely RGB (Red Green Blue), HSV (Hue, Saturation, Value), YUV, YIQ, YCbCr (Luminance, Chrominance) etc. are used to represent color images. Color image steganography can be done in any color space domain. In this paper color image steganography in RGB and YCbCr domain are compared. The secret information considered is grey scale image. Since RGB is the common method of representation, hiding secret information in this format is not secure.

Hemalatha S, U Dinesh Acharya, Renuka A et al.

Steganography is the science of invisible communication. The purpose of Steganography is to maintain secret communication between two parties. The secret information can be concealed in content such as image, audio, or video. This paper provides a novel image steganography technique to hide multiple secret images and keys in color cover image using Integer Wavelet Transform (IWT). There is no visual difference between the stego image and the cover image. The extracted secret images are also similar to the original secret images. Very good PSNR (Peak Signal to Noise Ratio) values are obtained for both stego and extracted secret images. The results are compared with the results of other techniques, where single image is hidden and it is found that the proposed technique is simple and gives better PSNR values than others.

Hemalatha S, U Dinesh Acharya, Renuka A et al.

Steganography is the art and science of covert communication. The secret information can be concealed in content such as image, audio, or video. This paper provides a novel image steganography technique to hide both image and key in color cover image using Discrete Wavelet Transform (DWT) and Integer Wavelet Transform (IWT). There is no visual difference between the stego image and the cover image. The extracted image is also similar to the secret image. This is proved by the high PSNR (Peak Signal to Noise Ratio), value for both stego and extracted secret image. The results are compared with the results of similar techniques and it is found that the proposed technique is simple and gives better PSNR values than others.