scholarly journals RIS-GAN: Explore Residual and Illumination with Generative Adversarial Networks for Shadow Removal

2020 ◽  
Vol 34 (07) ◽  
pp. 12829-12836 ◽  
Author(s):  
Ling Zhang ◽  
Chengjiang Long ◽  
Xiaolong Zhang ◽  
Chunxia Xiao
Keyword(s):  
Ground Truth ◽  
Superior Performance ◽  
Generative Adversarial Networks ◽  
Shadow Removal ◽  
Illumination Estimation ◽  
Adversarial Networks ◽  
The Arts ◽  
Benchmark Datasets ◽  
Coarse To Fine ◽  
Ground Truth Information

Residual images and illumination estimation have been proved very helpful in image enhancement. In this paper, we propose a general and novel framework RIS-GAN which explores residual and illumination with Generative Adversarial Networks for shadow removal. Combined with the coarse shadow-removal image, the estimated negative residual images and inverse illumination maps can be used to generate indirect shadow-removal images to refine the coarse shadow-removal result to the fine shadow-free image in a coarse-to-fine fashion. Three discriminators are designed to distinguish whether the predicted negative residual images, shadow-removal images, and the inverse illumination maps are real or fake jointly compared with the corresponding ground-truth information. To our best knowledge, we are the first one to explore residual and illumination for shadow removal. We evaluate our proposed method on two benchmark datasets, i.e., SRD and ISTD, and the extensive experiments demonstrate that our proposed method achieves the superior performance to state-of-the-arts, although we have no particular shadow-aware components designed in our generators.

scholarly journals Recurrent Relational Memory Network for Unsupervised Image Captioning

2020 ◽  
Author(s):  
Dan Guo ◽  
Yang Wang ◽  
Peipei Song ◽  
Meng Wang
Keyword(s):  
Generative Adversarial Networks ◽  
Relational Reasoning ◽  
Relational Memory ◽  
Image Captioning ◽  
Adversarial Networks ◽  
The Arts ◽  
Benchmark Datasets ◽  
Unsupervised Training ◽  
Visual Concepts ◽  
Memory Network

Unsupervised image captioning with no annotations is an emerging challenge in computer vision, where the existing arts usually adopt GAN (Generative Adversarial Networks) models. In this paper, we propose a novel memory-based network rather than GAN, named Recurrent Relational Memory Network (R2M). Unlike complicated and sensitive adversarial learning that non-ideally performs for long sentence generation, R2M implements a concepts-to-sentence memory translator through two-stage memory mechanisms: fusion and recurrent memories, correlating the relational reasoning between common visual concepts and the generated words for long periods. R2M encodes visual context through unsupervised training on images, while enabling the memory to learn from irrelevant textual corpus via supervised fashion. Our solution enjoys less learnable parameters and higher computational efficiency than GAN-based methods, which heavily bear parameter sensitivity. We experimentally validate the superiority of R2M than state-of-the-arts on all benchmark datasets.

scholarly journals RoCGAN: Robust Conditional GAN

2020 ◽  
Vol 128 (10-11) ◽  
pp. 2665-2683 ◽  
Author(s):  
Grigorios G. Chrysos ◽  
Jean Kossaifi ◽  
Stefanos Zafeiriou
Keyword(s):  
Large Scale ◽  
Real Data ◽  
Superior Performance ◽  
Target Space ◽  
Generative Adversarial Networks ◽  
Natural Scenes ◽  
Adversarial Networks ◽  
Target Manifold ◽  
The Face ◽  
Intense Noise

Abstract Conditional image generation lies at the heart of computer vision and conditional generative adversarial networks (cGAN) have recently become the method of choice for this task, owing to their superior performance. The focus so far has largely been on performance improvement, with little effort in making cGANs more robust to noise. However, the regression (of the generator) might lead to arbitrarily large errors in the output, which makes cGANs unreliable for real-world applications. In this work, we introduce a novel conditional GAN model, called RoCGAN, which leverages structure in the target space of the model to address the issue. Specifically, we augment the generator with an unsupervised pathway, which promotes the outputs of the generator to span the target manifold, even in the presence of intense noise. We prove that RoCGAN share similar theoretical properties as GAN and establish with both synthetic and real data the merits of our model. We perform a thorough experimental validation on large scale datasets for natural scenes and faces and observe that our model outperforms existing cGAN architectures by a large margin. We also empirically demonstrate the performance of our approach in the face of two types of noise (adversarial and Bernoulli).

scholarly journals Parallel Image Completion with Edge and Color Map

2019 ◽  
Vol 9 (18) ◽  
pp. 3856 ◽  
Author(s):  
Dan Zhao ◽  
Baolong Guo ◽  
Yunyi Yan
Keyword(s):  
Superior Performance ◽  
Generative Adversarial Networks ◽  
Image Completion ◽  
Significant Progress ◽  
Color Information ◽  
Parallel Edge ◽  
Adversarial Networks ◽  
Parallel Image ◽  
Latent Space ◽  
Image Edge

Over the last few years, image completion has made significant progress due to the generative adversarial networks (GANs) that are able to synthesize photorealistic contents. However, one of the main obstacles faced by many existing methods is that they often create blurry textures or distorted structures that are inconsistent with surrounding regions. The main reason is the ineffectiveness of disentangling style latent space implicitly from images. To address this problem, we develop a novel image completion framework called PIC-EC: parallel image completion networks with edge and color maps, which explicitly provides image edge and color information as the prior knowledge for image completion. The PIC-EC framework consists of the parallel edge and color generators followed by an image completion network. Specifically, the parallel paths generate edge and color maps for the missing region at the same time, and then the image completion network fills the missing region with fine details using the generated edge and color information as the priors. The proposed method was evaluated over CelebA-HQ and Paris StreetView datasets. Experimental results demonstrate that PIC-EC achieves superior performance on challenging cases with complex compositions and outperforms existing methods on evaluations of realism and accuracy, both quantitatively and qualitatively.

scholarly journals Multi-Instance Multi-Label Action Recognition and Localization Based on Spatio-Temporal Pre-Trimming for Untrimmed Videos

2020 ◽  
Vol 34 (07) ◽  
pp. 12886-12893
Author(s):  
Xiao-Yu Zhang ◽  
Haichao Shi ◽  
Changsheng Li ◽  
Peng Li
Keyword(s):  
Action Recognition ◽  
Human Action ◽  
Learning Problem ◽  
The Arts ◽  
Key Factor ◽  
Benchmark Datasets ◽  
Spatio Temporal ◽  
Weakly Supervised ◽  
Temporal Localization ◽  
Coarse To Fine

Weakly supervised action recognition and localization for untrimmed videos is a challenging problem with extensive applications. The overwhelming irrelevant background contents in untrimmed videos severely hamper effective identification of actions of interest. In this paper, we propose a novel multi-instance multi-label modeling network based on spatio-temporal pre-trimming to recognize actions and locate corresponding frames in untrimmed videos. Motivated by the fact that person is the key factor in a human action, we spatially and temporally segment each untrimmed video into person-centric clips with pose estimation and tracking techniques. Given the bag-of-instances structure associated with video-level labels, action recognition is naturally formulated as a multi-instance multi-label learning problem. The network is optimized iteratively with selective coarse-to-fine pre-trimming based on instance-label activation. After convergence, temporal localization is further achieved with local-global temporal class activation map. Extensive experiments are conducted on two benchmark datasets, i.e. THUMOS14 and ActivityNet1.3, and experimental results clearly corroborate the efficacy of our method when compared with the state-of-the-arts.

An anomaly detection method based on double encoder–decoder generative adversarial networks

2020 ◽  
Vol ahead-of-print (ahead-of-print) ◽  
Author(s):  
Hui Liu ◽  
Tinglong Tang ◽  
Jake Luo ◽  
Meng Zhao ◽  
Baole Zheng ◽  
...  
Keyword(s):  
Anomaly Detection ◽  
Latent Variables ◽  
Dimensional Space ◽  
Superior Performance ◽  
Generative Adversarial Networks ◽  
Training Process ◽  
Content Type ◽  
Detection Model ◽  
Adversarial Networks ◽  
Low Dimensional

Purpose This study aims to address the challenge of training a detection model for the robot to detect the abnormal samples in the industrial environment, while abnormal patterns are very rare under this condition. Design/methodology/approach The authors propose a new model with double encoder–decoder (DED) generative adversarial networks to detect anomalies when the model is trained without any abnormal patterns. The DED approach is used to map high-dimensional input images to a low-dimensional space, through which the latent variables are obtained. Minimizing the change in the latent variables during the training process helps the model learn the data distribution. Anomaly detection is achieved by calculating the distance between two low-dimensional vectors obtained from two encoders. Findings The proposed method has better accuracy and F1 score when compared with traditional anomaly detection models. Originality/value A new architecture with a DED pipeline is designed to capture the distribution of images in the training process so that anomalous samples are accurately identified. A new weight function is introduced to control the proportion of losses in the encoding reconstruction and adversarial phases to achieve better results. An anomaly detection model is proposed to achieve superior performance against prior state-of-the-art approaches.

scholarly journals Deep image synthesis from intuitive user input: A review and perspectives

2021 ◽  
Vol 8 (1) ◽  
pp. 3-31
Author(s):  
Yuan Xue ◽  
Yuan-Chen Guo ◽  
Han Zhang ◽  
Tao Xu ◽  
Song-Hai Zhang ◽  
...  
Keyword(s):  
Image Synthesis ◽  
Generative Models ◽  
Generative Adversarial Networks ◽  
Image Generation ◽  
Art And Design ◽  
User Input ◽  
Adversarial Networks ◽  
Benchmark Datasets ◽  
Deep Image ◽  
Realistic Images

AbstractIn many applications of computer graphics, art, and design, it is desirable for a user to provide intuitive non-image input, such as text, sketch, stroke, graph, or layout, and have a computer system automatically generate photo-realistic images according to that input. While classically, works that allow such automatic image content generation have followed a framework of image retrieval and composition, recent advances in deep generative models such as generative adversarial networks (GANs), variational autoencoders (VAEs), and flow-based methods have enabled more powerful and versatile image generation approaches. This paper reviews recent works for image synthesis given intuitive user input, covering advances in input versatility, image generation methodology, benchmark datasets, and evaluation metrics. This motivates new perspectives on input representation and interactivity, cross fertilization between major image generation paradigms, and evaluation and comparison of generation methods.

scholarly journals Deep Convolutional Sum-Product Networks

2019 ◽  
Vol 33 ◽  
pp. 3248-3255
Author(s):  
Cory J. Butz ◽  
Jhonatan S. Oliveira ◽  
André E. Dos Santos ◽  
André L. Teixeira
Keyword(s):  
Probabilistic Reasoning ◽  
Mean Squared Error ◽  
Salient Feature ◽  
Generative Adversarial Networks ◽  
Image Sampling ◽  
Adversarial Networks ◽  
Squared Error ◽  
Benchmark Datasets ◽  
Most Probable Explanation ◽  
Marginal Inference

We give conditions under which convolutional neural networks (CNNs) define valid sum-product networks (SPNs). One subclass, called convolutional SPNs (CSPNs), can be implemented using tensors, but also can suffer from being too shallow. Fortunately, tensors can be augmented while maintaining valid SPNs. This yields a larger subclass of CNNs, which we call deep convolutional SPNs (DCSPNs), where the convolutional and sum-pooling layers form rich directed acyclic graph structures. One salient feature of DCSPNs is that they are a rigorous probabilistic model. As such, they can exploit multiple kinds of probabilistic reasoning, including marginal inference and most probable explanation (MPE) inference. This allows an alternative method for learning DCSPNs using vectorized differentiable MPE, which plays a similar role to the generator in generative adversarial networks (GANs). Image sampling is yet another application demonstrating the robustness of DCSPNs. Our preliminary results on image sampling are encouraging, since the DCSPN sampled images exhibit variability. Experiments on image completion show that DCSPNs significantly outperform competing methods by achieving several state-of-the-art mean squared error (MSE) scores in both left-completion and bottom-completion in benchmark datasets.

scholarly journals Classification and pattern extraction of incidents: a deep learning-based approach

2022 ◽  
Author(s):  
Sobhan Sarkar ◽  
Sammangi Vinay ◽  
Chawki Djeddi ◽  
J. Maiti
Keyword(s):  
Superior Performance ◽  
Hidden Information ◽  
Text Data ◽  
Moment Estimation ◽  
The Arts ◽  
Proposed Model ◽  
Benchmark Datasets ◽  
Unexplored Area ◽  
Key Terms

AbstractClassifying or predicting occupational incidents using both structured and unstructured (text) data are an unexplored area of research. Unstructured texts, i.e., incident narratives are often unutilized or underutilized. Besides the explicit information, there exist a large amount of hidden information present in a dataset, which cannot be explored by the traditional machine learning (ML) algorithms. There is a scarcity of studies that reveal the use of deep neural networks (DNNs) in the domain of incident prediction, and its parameter optimization for achieving better prediction power. To address these issues, initially, key terms are extracted from the unstructured texts using LDA-based topic modeling. Then, these key terms are added with the predictor categories to form the feature vector, which is further processed for noise reduction and fed to the adaptive moment estimation (ADAM)-based DNN (i.e., ADNN) for classification, as ADAM is superior to GD, SGD, and RMSProp. To evaluate the effectiveness of our proposed method, a comparative study has been conducted using some state-of-the-arts on five benchmark datasets. Moreover, a case study of an integrated steel plant in India has been demonstrated for the validation of the proposed model. Experimental results reveal that ADNN produces superior performance than others in terms of accuracy. Therefore, the present study offers a robust methodological guide that enables us to handle the issues of unstructured data and hidden information for developing a predictive model.

Sign in / Sign up

Export Citation Format

Share Document