Advances in technology have led to the development of methods that can create desired visual multimedia. In particular, image generation using deep learning has been extensively studied across diverse fields. In comparison, video generation, especially on conditional inputs, remains a challenging and less explored area. To narrow this gap, we aim to train our model to produce a video corresponding to a given text description. We propose a novel training framework, Text-to-Image-to-Video Generative Adversarial Network (TiVGAN), which evolves frame-by-frame and finally produces a full-length video. In the first phase, we focus on creating a high-quality single video frame while learning the relationship between the text and an image. As the steps proceed, our model is trained gradually on more number of consecutive frames.This step-by-step learning process helps stabilize the training and enables the creation of high-resolution video based on conditional text descriptions. Qualitative and quantitative experimental results on various datasets demonstrate the effectiveness of the proposed method.
We propose a novel pose-guided appearance transfer network for transferring a given reference appearance to a target pose in unprecedented image resolution (1024 * 1024), given respectively an image of the reference and target person. No 3D model is used. Instead, our network utilizes dense local descriptors including local perceptual loss and local discriminators to refine details, which is trained progressively in a coarse-to-fine manner to produce the high-resolution output to faithfully preserve complex appearance of garment textures and geometry, while hallucinating seamlessly the transferred appearances including those with dis-occlusion. Our progressive encoder-decoder architecture can learn the reference appearance inherent in the input image at multiple scales. Extensive experimental results on the Human3.6M dataset, the DeepFashion dataset, and our dataset collected from YouTube show that our model produces high-quality images, which can be further utilized in useful applications such as garment transfer between people and pose-guided human video generation.
In this paper, we present a novel network for high resolution video generation. Our network uses ideas from Wasserstein GANs by enforcing k-Lipschitz constraint on the loss term and Conditional GANs using class labels for training and testing. We present Generator and Discriminator network layerwise details along with the combined network architecture, optimization details and algorithm used in this work. Our network uses a combination of two loss terms: mean square pixel loss and an adversarial loss. The datasets used for training and testing our network are UCF101, Golf and Aeroplane Datasets. Using Inception Score and Fr\'echet Inception Distance as the evaluation metrics, our network outperforms previous state of the art networks on unsupervised video generation.
Fake portrait video generation techniques have been posing a new threat to the society with photorealistic deep fakes for political propaganda, celebrity imitation, forged evidences, and other identity related manipulations. Following these generation techniques, some detection approaches have also been proved useful due to their high classification accuracy. Nevertheless, almost no effort was spent to track down the source of deep fakes. We propose an approach not only to separate deep fakes from real videos, but also to discover the specific generative model behind a deep fake. Some pure deep learning based approaches try to classify deep fakes using CNNs where they actually learn the residuals of the generator. We believe that these residuals contain more information and we can reveal these manipulation artifacts by disentangling them with biological signals. Our key observation yields that the spatiotemporal patterns in biological signals can be conceived as a representative projection of residuals. To justify this observation, we extract PPG cells from real and fake videos and feed these to a state-of-the-art classification network for detecting the generative model per video. Our results indicate that our approach can detect fake videos with 97.29% accuracy, and the source model with 93.39% accuracy.
This paper proposes a novel model for video generation and especially makes the attempt to deal with the problem of video generation from text descriptions, i.e., synthesizing realistic videos conditioned on given texts. Existing video generation methods cannot be easily adapted to handle this task well, due to the frame discontinuity issue and their text-free generation schemes. To address these problems, we propose a recurrent deconvolutional generative adversarial network (RD-GAN), which includes a recurrent deconvolutional network (RDN) as the generator and a 3D convolutional neural network (3D-CNN) as the discriminator. The RDN is a deconvolutional version of conventional recurrent neural network, which can well model the long-range temporal dependency of generated video frames and make good use of conditional information. The proposed model can be jointly trained by pushing the RDN to generate realistic videos so that the 3D-CNN cannot distinguish them from real ones. We apply the proposed RD-GAN to a series of tasks including conventional video generation, conditional video generation, video prediction and video classification, and demonstrate its effectiveness by achieving well performance.
This paper presents a novel end-to-end dynamic time-lapse video generation framework, named DTVNet, to generate diversified time-lapse videos from a single landscape image, which are conditioned on normalized motion vectors. The proposed DTVNet consists of two submodules: \emph{Optical Flow Encoder} (OFE) and \emph{Dynamic Video Generator} (DVG). The OFE maps a sequence of optical flow maps to a \emph{normalized motion vector} that encodes the motion information inside the generated video. The DVG contains motion and content streams that learn from the motion vector and the single image respectively, as well as an encoder and a decoder to learn shared content features and construct video frames with corresponding motion respectively. Specifically, the \emph{motion stream} introduces multiple \emph{adaptive instance normalization} (AdaIN) layers to integrate multi-level motion information that are processed by linear layers. In the testing stage, videos with the same content but various motion information can be generated by different \emph{normalized motion vectors} based on only one input image. We further conduct experiments on Sky Time-lapse dataset, and the results demonstrate the superiority of our approach over the state-of-the-art methods for generating high-quality and dynamic videos, as well as the variety for generating videos with various motion information.
A natural approach to generative modeling of videos is to represent them as a composition of moving objects. Recent works model a set of 2D sprites over a slowly-varying background, but without considering the underlying 3D scene that gives rise to them. We instead propose to model a video as the view seen while moving through a scene with multiple 3D objects and a 3D background. Our model is trained from monocular videos without any supervision, yet learns to generate coherent 3D scenes containing several moving objects. We conduct detailed experiments on two datasets, going beyond the visual complexity supported by state-of-the-art generative approaches. We evaluate our method on depth-prediction and 3D object detection -- tasks which cannot be addressed by those earlier works -- and show it out-performs them even on 2D instance segmentation and tracking.
Novel-View Human Action Synthesis aims to synthesize the movement of a body from a virtual viewpoint, given a video from a real viewpoint. We present a novel 3D reasoning to synthesize the target viewpoint. We first estimate the 3D mesh of the target body and transfer the rough textures from the 2D images to the mesh. As this transfer may generate sparse textures on the mesh due to frame resolution or occlusions. We produce a semi-dense textured mesh by propagating the transferred textures both locally, within local geodesic neighborhoods, and globally, across symmetric semantic parts. Next, we introduce a context-based generator to learn how to correct and complete the residual appearance information. This allows the network to independently focus on learning the foreground and background synthesis tasks. We validate the proposed solution on the public NTU RGB+D dataset. The code and resources are available at https://bit.ly/36u3h4K.
Generating long-range skeleton-based human actions has been a challenging problem since small deviations of one frame can cause a malformed action sequence. Most existing methods borrow ideas from video generation, which naively treat skeleton nodes/joints as pixels of images without considering the rich inter-frame and intra-frame structure information, leading to potential distorted actions. Graph convolutional networks (GCNs) is a promising way to leverage structure information to learn structure representations. However, directly adopting GCNs to tackle such continuous action sequences both in spatial and temporal spaces is challenging as the action graph could be huge. To overcome this issue, we propose a variant of GCNs to leverage the powerful self-attention mechanism to adaptively sparsify a complete action graph in the temporal space. Our method could dynamically attend to important past frames and construct a sparse graph to apply in the GCN framework, well-capturing the structure information in action sequences. Extensive experimental results demonstrate the superiority of our method on two standard human action datasets compared with existing methods.
Videos of actions are complex signals containing rich compositional structure in space and time. Current video generation methods lack the ability to condition the generation on multiple coordinated and potentially simultaneous timed actions. To address this challenge, we propose to represent the actions in a graph structure called Action Graph and present the new ``Action Graph To Video'' synthesis task. Our generative model for this task (AG2Vid) disentangles motion and appearance features, and by incorporating a scheduling mechanism for actions facilitates a timely and coordinated video generation. We train and evaluate AG2Vid on the CATER and Something-Something V2 datasets, and show that the resulting videos have better visual quality and semantic consistency compared to baselines. Finally, our model demonstrates zero-shot abilities by synthesizing novel compositions of the learned actions. For code and pretrained models, see the project page https://roeiherz.github.io/AG2Video
We present a novel system that gets as an input video frames of a musician playing the piano and generates the music for that video. Generation of music from visual cues is a challenging problem and it is not clear whether it is an attainable goal at all. Our main aim in this work is to explore the plausibility of such a transformation and to identify cues and components able to carry the association of sounds with visual events. To achieve the transformation we built a full pipeline named `\textit{Audeo}' containing three components. We first translate the video frames of the keyboard and the musician hand movements into raw mechanical musical symbolic representation Piano-Roll (Roll) for each video frame which represents the keys pressed at each time step. We then adapt the Roll to be amenable for audio synthesis by including temporal correlations. This step turns out to be critical for meaningful audio generation. As a last step, we implement Midi synthesizers to generate realistic music. \textit{Audeo} converts video to audio smoothly and clearly with only a few setup constraints. We evaluate \textit{Audeo} on `in the wild' piano performance videos and obtain that their generated music is of reasonable audio quality and can be successfully recognized with high precision by popular music identification software.
We attempt to recognize and track lyric words in lyric videos. Lyric video is a music video showing the lyric words of a song. The main characteristic of lyric videos is that the lyric words are shown at frames synchronously with the music. The difficulty of recognizing and tracking the lyric words is that (1) the words are often decorated and geometrically distorted and (2) the words move arbitrarily and drastically in the video frame. The purpose of this paper is to analyze the motion of the lyric words in lyric videos, as the first step of automatic lyric video generation. In order to analyze the motion of lyric words, we first apply a state-of-the-art scene text detector and recognizer to each video frame. Then, lyric-frame matching is performed to establish the optimal correspondence between lyric words and the frames. After fixing the motion trajectories of individual lyric words from correspondence, we analyze the trajectories of the lyric words by k-medoids clustering and dynamic time warping (DTW).
We consider the task of generating diverse and novel videos from a single video sample. Recently, new hierarchical patch-GAN based approaches were proposed for generating diverse images, given only a single sample at training time. Moving to videos, these approaches fail to generate diverse samples, and often collapse into generating samples similar to the training video. We introduce a novel patch-based variational autoencoder (VAE) which allows for a much greater diversity in generation. Using this tool, a new hierarchical video generation scheme is constructed: at coarse scales, our patch-VAE is employed, ensuring samples are of high diversity. Subsequently, at finer scales, a patch-GAN renders the fine details, resulting in high quality videos. Our experiments show that the proposed method produces diverse samples in both the image domain, and the more challenging video domain.
The video generation task can be formulated as a prediction of future video frames given some past frames. Recent generative models for videos face the problem of high computational requirements. Some models require up to 512 Tensor Processing Units for parallel training. In this work, we address this problem via modeling the dynamics in a latent space. After the transformation of frames into the latent space, our model predicts latent representation for the next frames in an autoregressive manner. We demonstrate the performance of our approach on BAIR Robot Pushing and Kinetics-600 datasets. The approach tends to reduce requirements to 8 Graphical Processing Units for training the models while maintaining comparable generation quality.
As the GAN-based face image and video generation techniques, widely known as DeepFakes, have become more and more matured and realistic, there comes a pressing and urgent demand for effective DeepFakes detectors. Motivated by the fact that remote visual photoplethysmography (PPG) is made possible by monitoring the minuscule periodic changes of skin color due to blood pumping through the face, we conjecture that normal heartbeat rhythms found in the real face videos will be disrupted or even entirely broken in a DeepFake video, making it a potentially powerful indicator for DeepFake detection. In this work, we propose DeepRhythm, a DeepFake detection technique that exposes DeepFakes by monitoring the heartbeat rhythms. DeepRhythm utilizes dual-spatial-temporal attention to adapt to dynamically changing face and fake types. Extensive experiments on FaceForensics++ and DFDC-preview datasets have confirmed our conjecture and demonstrated not only the effectiveness, but also the generalization capability of \emph{DeepRhythm} over different datasets by various DeepFakes generation techniques and multifarious challenging degradations.
Though deep learning based scene text detection has achieved great progress, well-trained detectors suffer from severe performance degradation for different domains. In general, a tremendous amount of data is indispensable to train the detector in the target domain. However, data collection and annotation are expensive and time-consuming. To address this problem, we propose a self-training framework to automatically mine hard examples with pseudo-labels from unannotated videos or images. To reduce the noise of hard examples, a novel text mining module is implemented based on the fusion of detection and tracking results. Then, an image-to-video generation method is designed for the tasks that videos are unavailable and only images can be used. Experimental results on standard benchmarks, including ICDAR2015, MSRA-TD500, ICDAR2017 MLT, demonstrate the effectiveness of our self-training method. The simple Mask R-CNN adapted with self-training and fine-tuned on real data can achieve comparable or even superior results with the state-of-the-art methods.
Ensemble smoothers are among the most successful and efficient techniques currently available for history matching. However, because these methods rely on Gaussian assumptions, their performance is severely degraded when the prior geology is described in terms of complex facies distributions. Inspired by the impressive results obtained by deep generative networks in areas such as image and video generation, we started an investigation focused on the use of autoencoders networks to construct a continuous parameterization for facies models. In our previous publication, we combined a convolutional variational autoencoder (VAE) with the ensemble smoother with multiple data assimilation (ES-MDA) for history matching production data in models generated with multiple-point geostatistics. Despite the good results reported in our previous publication, a major limitation of the designed parameterization is the fact that it does not allow applying distance-based localization during the ensemble smoother update, which limits its application in large-scale problems. The present work is a continuation of this research project focusing in two aspects: firstly, we benchmark seven different formulations, including VAE, generative adversarial network (GAN), Wasserstein GAN, variational auto-encoding GAN, principal component analysis (PCA) with cycle GAN, PCA with transfer style network and VAE with style loss. These formulations are tested in a synthetic history matching problem with channelized facies. Secondly, we propose two strategies to allow the use of distance-based localization with the deep learning parameterizations.
Over the years, performance evaluation has become essential in computer vision, enabling tangible progress in many sub-fields. While talking-head video generation has become an emerging research topic, existing evaluations on this topic present many limitations. For example, most approaches use human subjects (e.g., via Amazon MTurk) to evaluate their research claims directly. This subjective evaluation is cumbersome, unreproducible, and may impend the evolution of new research. In this work, we present a carefully-designed benchmark for evaluating talking-head video generation with standardized dataset pre-processing strategies. As for evaluation, we either propose new metrics or select the most appropriate ones to evaluate results in what we consider as desired properties for a good talking-head video, namely, identity preserving, lip synchronization, high video quality, and natural-spontaneous motion. By conducting a thoughtful analysis across several state-of-the-art talking-head generation approaches, we aim to uncover the merits and drawbacks of current methods and point out promising directions for future work. All the evaluation code is available at: https://github.com/lelechen63/talking-head-generation-survey.
Video generation models have become increasingly popular in the last few years, however the standard 2D architectures used today lack natural spatio-temporal modelling capabilities. In this paper, we present a network architecture for video generation that models spatio-temporal consistency without resorting to costly 3D architectures. The architecture facilitates information exchange between neighboring time points, which improves the temporal consistency of both the high level structure as well as the low-level details of the generated frames. The approach achieves state-of-the-art quantitative performance, as measured by the inception score on the UCF-101 dataset as well as better qualitative results. We also introduce a new quantitative measure (S3) that uses downstream tasks for evaluation. Moreover, we present a new multi-label dataset MaisToy, which enables us to evaluate the generalization of the model.
Recent breakthroughs in adversarial generative modeling have led to models capable of producing video samples of high quality, even on large and complex datasets of real-world video. In this work, we focus on the task of video prediction, where given a sequence of frames extracted from a video, the goal is to generate a plausible future sequence. We first improve the state of the art by performing a systematic empirical study of discriminator decompositions and proposing an architecture that yields faster convergence and higher performance than previous approaches. We then analyze recurrent units in the generator, and propose a novel recurrent unit which transforms its past hidden state according to predicted motion-like features, and refines it to handle dis-occlusions, scene changes and other complex behavior. We show that this recurrent unit consistently outperforms previous designs. Our final model leads to a leap in the state-of-the-art performance, obtaining a test set Frechet Video Distance of 25.7, down from 69.2, on the large-scale Kinetics-600 dataset.