- 标题
- 摘要
- 关键词
- 实验方案
- 产品
-
[IEEE 2018 24th International Conference on Pattern Recognition (ICPR) - Beijing (2018.8.20-2018.8.24)] 2018 24th International Conference on Pattern Recognition (ICPR) - An Extensive Study of Cycle-Consistent Generative Networks for Image-to-Image Translation
摘要： Image-to-image translation between different domains has been an important research direction, with the aim of arbitrarily manipulating the source image content to become similar to a target image. Recently, cycle-consistent generative network (CycleGAN) has become a fundamental approach for general-purpose image-to-image translation, while almost no work has examined what factors may influence its performance. To provide more insights, we propose two new models roughly based on CycleGAN, namely LongCycleGAN and NestCycleGAN. First, LongCycleGAN cascades several generators to perform the domain translation in a long cycle. It shows the benefit of stacking more generators on the generation quality. In addition to the long cycle, NestCycleGAN develops new inner cycles to bridge intermediate generators directly, which can help constrain the unsupervised mappings. In the experiments, we conduct qualitative and quantitative comparisons for tasks including photo?label, photo?sketch, and photo colorization. The quantitative and qualitative results demonstrate the effectiveness of our two proposed models.
关键词： CycleGAN,NestCycleGAN,unsupervised learning,image-to-image translation,cycle-consistent generative networks,LongCycleGAN
更新于2025-09-23 15:22:29
-
Single-Molecule Imaging of mRNA Localization and Regulation during the Integrated Stress Response
摘要： Biological phase transitions form membrane-less organelles that generate distinct cellular environments. How molecules are partitioned between these compartments and the surrounding cellular space and the functional consequence of this localization is not well understood. Here, we report the localization of mRNA to stress granules (SGs) and processing bodies (PBs) and its effect on translation and degradation during the integrated stress response. Using single mRNA imaging in living human cells, we find that the interactions of mRNAs with SGs and PBs have different dynamics, very few mRNAs directly move between SGs and PBs, and that specific RNA-binding proteins can anchor mRNAs within these compartments. During recovery from stress, we show that mRNAs that were within SGs and PBs are translated and degraded at similar rates as their cytosolic counterparts. Our work provides a framework for using single-molecule measurements to directly investigate the molecular mechanisms of phase-separated compartments within their cellular environment.
关键词： P-bodies,integrated stress response,degradation,stress granules,LARP1,mRNA localization,single-molecule imaging,translation
更新于2025-09-23 15:22:29
-
[IEEE 2018 24th International Conference on Pattern Recognition (ICPR) - Beijing (2018.8.20-2018.8.24)] 2018 24th International Conference on Pattern Recognition (ICPR) - Unsupervised Multi-Domain Image Translation with Domain-Specific Encoders/Decoders
摘要： Unsupervised Image-to-Image Translation achieves spectacularly advanced developments nowadays. However, recent approaches mainly focus on one model with two domains, which may face heavy burdens with the large cost of training time and the huge model parameters, under such a requirement that n (n>2) domains are freely transferred to each other in a general setting. To address this problem, we propose a novel and unified framework named Domain-Bank, which consists of a globally shared auto-encoder and n domain-specific encoders/decoders, assuming that there is a universal shared-latent space can be projected. Thus, we not only reduce the parameters of the model but also have a huge reduction of the time budgets. Besides the high efficiency, we show the comparable (or even better) image translation results over state-of-the-arts on various challenging unsupervised image translation tasks, including face image translation and painting style translation. We also apply the proposed framework to the domain adaptation task and achieve state-of-the-art performance on digit benchmark datasets.
关键词： Shared-Latent Space,Unsupervised Image-to-Image Translation,Generative Adversarial Networks,Variational Autoencoders,Domain-Bank,Multi-Domain
更新于2025-09-23 15:22:29
-
Image-Translation-Based Road Marking Extraction From Mobile Laser Point Clouds
摘要： Road markings are one of the most important safety elements in a road network, and they play a critical role in traffic safety. However, the automatic extraction of road markings remains a technical challenge in the fields of smart city construction and automatic driving. This paper presents an image-translation-based method of obtaining the 3D vectors of typical road markings from mobile laser point clouds. First, ground roughness is used as a criterion to extract ground points based on the topological relationship of adjacent scan lines, and the feature images of a road surface are generated using the adapted inverse distance weighted method. Second, by comparing objective functions based on the pix2pix framework, a finely adjusted image-to-image translation model named P2P_L1 is proposed for the segmentation of road markings. The proposed model outperforms the advanced DeepLab V3+ network in terms of precision, F1-score, and mean Intersection over Union indicators in the comparative segmentation results of ten types of road markings in the Shenzhen test area. Third, methods such as node averaging and optimized iterative closest point are developed for the 3D vectorization of road markings. This study presents a new approach for the automatic extraction of road markings to provide effective technical support for the construction of smart cities.
关键词： image translation,segmentation,road marking,Conditional generative adversarial nets (cGANs),laser point cloud
更新于2025-09-23 15:19:57
-
[IEEE 2019 IEEE Intelligent Vehicles Symposium (IV) - Paris, France (2019.6.9-2019.6.12)] 2019 IEEE Intelligent Vehicles Symposium (IV) - DeLiO: Decoupled LiDAR Odometry
摘要： Most LiDAR odometry algorithms estimate the transformation between two consecutive frames by estimating the rotation and translation in an intervening fashion. In this paper, we propose our Decoupled LiDAR Odometry (DeLiO), which – for the first time – decouples the rotation estimation completely from the translation estimation. In particular, the rotation is estimated by extracting the surface normals from the input point clouds and tracking their characteristic pattern on a unit sphere. Using this rotation the point clouds are unrotated so that the underlying transformation is pure translation, which can be easily estimated using a line cloud approach. An evaluation is performed on the KITTI dataset and the results are compared against state-of-the-art algorithms.
关键词： decoupled rotation and translation estimation,KITTI dataset,line cloud approach,LiDAR odometry,unit sphere,surface normals
更新于2025-09-12 10:27:22
-
Translate SAR Data into Optical Image Using IHS and Wavelet Transform Integrated Fusion
摘要： Although synthetic aperture radar (SAR) sensors function well at all times and under all weather conditions, the images they produce are not intuitively straightforward. A novel idea based on data fusion is introduced to translate SAR data into optical image in this paper. The proposed SAR-optical image translation is implemented using an intensity–hue–saturation (IHS) and wavelet transform integrated fusion (IHSW), so as to preserve as much as spatial details from SAR data, and minimize the spectral distortion of translated output. COSMO-SkyMed and ENVISAT-ASAR images are translated into optical images with the fusion of Landsat TM images, and the fusion results are compared with some conventional fusion methods, as well as the texture synthesis approach. Quality assessment of different fused outputs is carried out by using six quality indices. Visual and statistical comparisons of the ?nal results indicate that the proposed approach achieves an effective translation from SAR to optical image and is superior to texture synthesis-based algorithm in terms of preserving spatial and spectral information. The proposed translation technique presents an alternative to improve the interpretability of SAR images.
关键词： Image fusion,Synthetic aperture radar (SAR),Image translation,Wavelet transform
更新于2025-09-09 09:28:46
-
[IEEE 2018 IEEE 7th Global Conference on Consumer Electronics (GCCE) - Nara, Japan (2018.10.9-2018.10.12)] 2018 IEEE 7th Global Conference on Consumer Electronics (GCCE) - Eye Gaze Correction Using Generative Adversarial Networks
摘要： Eye gaze correction is an important topic in video teleconference and video chart in order to keep the eye contact. In this paper, we propose to use a generative adversarial networks for eye gaze correction. We use pairs of front facial image (idea camera setting) and real facial image (real camera setting) to training the network. By using the trained network, we can generate a gaze corrected facial image (front facial image) for any real facial image. Experiments demonstrated the effectiveness of our proposed method.
关键词： Generative Adversarial Net(GAN),image-to-image translation,deep learning,gaze correction,Conditional GAN
更新于2025-09-04 15:30:14
-
[IEEE 2018 IEEE 7th Global Conference on Consumer Electronics (GCCE) - Nara, Japan (2018.10.9-2018.10.12)] 2018 IEEE 7th Global Conference on Consumer Electronics (GCCE) - Automatic Generation of Facial Expression Using Generative Adversarial Nets
摘要： With the spread of digital cameras, smart phones, and SNS, the number facial images of people have increased. Facial expression generation from a single facial image has been widely applied to the fields of entertainment and social communication. Many approaches that apply machine learning techniques have been developed. In our previous study, we developed a makeup simulator system. However, this system is incapable of changing the impression of a cosmetic face based on changes in facial expression; in addition, another challenge is that the user cannot see the impression of makeup dynamically and objectively. Therefore, in this study, we generate static facial expression images from a natural (expressionless) image by using generative adversarial networks, which is critical to the research on dynamic facial expression change. Our experimental results demonstrate that our approach achieves the best expression image.
关键词： Generative Adversarial Nets,image,Image-to-Image Translation with Conditional Adversarial Networks,facial expression
更新于2025-09-04 15:30:14
-
SAR Automatic Target Recognition Using a Roto-Translational Invariant Wavelet-Scattering Convolution Network
摘要： The algorithm of synthetic aperture radar (SAR) for automatic target recognition consists of two stages: feature extraction and classification. The quality of extracted features has significant impacts on the final classification performance. This paper presents a SAR automatic target classification method based on the wavelet-scattering convolution network. By introducing a deep scattering convolution network with complex wavelet filters over spatial and angular variables, robust feature representations can be extracted across various scales and angles without training data. Conventional dimension reduction and a support vector machine classifier are followed to complete the classification task. The proposed method is then tested on the moving and stationary target acquisition and recognition (MSTAR) benchmark data set and achieves an average accuracy of 97.63% on the classification of ten-class targets without data augmentation.
关键词： automatic target classification (ATR),wavelet transform,scattering convolution network,roto-translation invariance,synthetic aperture radar
更新于2025-09-04 15:30:14