Video super-resolution (VSR) is a fundamental and challenging task in computer vision. Many of the existing VSR works focus on how to effectively align neighboring frames to better incorporate temporal information, while little work is devoted to the important subsequent step of inter-frame information fusion, and the existing methods on frame fusion have shortcomings such as not being able to make full use of spatio-temporal information. In this work, we propose a Frame-by-frame Feedback Fusion Network (FFFN) for VSR tasks. By applying the feedback learning mechanism commonly existing in the human cognitive system to the frame fusion stage, FFFN can refine low-level representation of the fused frames with high-level information in a coarse-to-fine manner. Specifically, after the neighboring frames are aligned, we first rearrange them from near to far according to the distance from the reference frame in the temporal space, and then feed them one-by-one into a proposed recurrent structure called Feedback Fusion Module (FFM), which is then able to iteratively generate high-level representation of the fused frames with several Feature Refinement Groups (FRGs) and feedback connections. Finally, we design a Dual-path Residual Reconstruction Module (DRRM) to reconstruct the final high-resolution image. The proposed FFFN comes with a strong frame fusion and reconstruction ability, and extensive experiments on several benchmark data sets show that it achieves favorable performance against state-of-the-art methods.

Despite the significant progress in unsupervised domain adaptation (UDA), the performance of UDA methods is still far inferior to that of the fully supervised ones. In practical scenarios, it is usually feasible to acquire labels on a small portion of the target data through active learning (AL), which aims to train an effective model with as few queried instances as possible. However, due to the domain shift, the instances selected by existing AL algorithms can be uninformative, redundant, or outlying. To address this issue, we propose a novel approach, namely, clustering environment-aware learning (CEAL), for active domain adaptation (ADA). CEAL selects potentially the most valuable instances under domain shift by exploring the informativeness and representativeness of target samples in a clustering environment-aware manner. Specifically, for the informativeness, we not only leverage the knowledge of individual points but also their nearby neighbors, by measuring the proposed clustering environment aware informativeness score (CEAIS), thus ensuring that the selected samples are highly informative. For the representativeness, we design two schemes called point distance release (PDR) and informativeness score difference exclusion (ISDE) to guarantee the diversity and validity of the selected samples. Furthermore, we fully utilize the large amount of unlabeled data from target domain via pseudo labeling and adopt information maximization to improve the reliability of the target pseudo labels, thereby further improving the performance of the model. The effectiveness of our method is empirically verified on various benchmark datasets against recent state-of-the-art algorithms.

We report the first (to the best of our knowledge) high-power, low-coherence Nd:glass laser delivering kilojoule pulses with a coherent time of 249 fs and a bandwidth of 13 nm, achieving the 63%-efficiency second-harmonic conversion of the large-aperture low-coherence pulse and good beam smoothing effect. It provides a new type of laser driver for laser plasma interaction and high energy density physics research.

Deep learning approaches for multi-label Chest X-ray (CXR) images classification usually require large-scale datasets. However, acquiring such datasets with full annotations is costly, time-consuming, and prone to noisy labels. Therefore, we introduce a weakly supervised learning problem called Single Positive Multi-label Learning (SPML) into CXR images classification (abbreviated as SPML-CXR), in which only one positive label is annotated per image. A simple solution to SPML-CXR problem is to assume that all the unannotated pathological labels are negative, however, it might introduce false negative labels and decrease the model performance. To this end, we present a Multi-level Pseudo-label Consistency (MPC) framework for SPML-CXR. First, inspired by the pseudo-labeling and consistency regularization in semi-supervised learning, we construct a weak-to-strong consistency framework, where the model prediction on weakly-augmented image is treated as the pseudo label for supervising the model prediction on a strongly-augmented version of the same image, and define an Image-level Perturbation-based Consistency (IPC) regularization to recover the potential mislabeled positive labels. Besides, we incorporate Random Elastic Deformation (RED) as an additional strong augmentation to enhance the perturbation. Second, aiming to expand the perturbation space, we design a perturbation stream to the consistency framework at the feature-level and introduce a Feature-level Perturbation-based Consistency (FPC) regularization as a supplement. Third, we design a Transformer-based encoder module to explore the sample relationship within each mini-batch by a Batch-level Transformer-based Correlation (BTC) regularization. Extensive experiments on the CheXpert and MIMIC-CXR datasets have shown the effectiveness of our MPC framework for solving the SPML-CXR problem.

Palmprint recognition provides a potential solution for noninvasive personal authentication due to its excellent contactless property and user-security, and it has attracted tremendous research interest in recent years. However, most existing methods focus on intraspectral palmprint recognition, which requires gallery and probe images to be captured under similar illumination, and thus significantly limit its practical applications in open environments with variant illuminations. In this study, we present a spectrum-invariant feature learning method for cross-spectral palmprint recognition to address the problem that gallery and probe samples are captured under different spectra. First, the blockwise direction-based ordinal measure vectors are formed to represent the intrinsic information of palmprint images. Then, a unified feature projection is jointly learned to map two different spectra of palmprint images into a common feature space, in which the different spectral features have enhanced discriminative power by enlarging their variances while the intraclass features learned from different spectral images are similar. The proposed method can be easily extended to seek the unified spectrum-invariant representation of multiple spectral palmprint images, making it feasible to perform palmprint recognition crossing one spectrum to multiple spectra. Experimental results on two multispectral palmprint image databases demonstrate the promising effectiveness of the proposed method on cross-spectral palmprint recognition.

Recently, Point-MAE has extended Masked Autoencoders (MAE) to point clouds for 3D self-supervised learning, which however faces two problems: (1) the shape similarity between the masked point cloud and original point cloud is high, and (2) the pretext task of reconstructing the original point cloud is straightforward which fails to compel the network to learn deep representative features. In this paper, we tackle these problems by proposing a PatchMixing strategy and a teacher-student training framework. First, with PatchMixing, we mix selected point patches of multiple point clouds and attempt to infer the object information from the resulting mixed point cloud. Due to the interference of other objects, the task is challenging but facilitates representation learning. Second, rather than directly restoring the original point cloud, we propose a novel pretext task that involves a two-branch teacher model and a student model. These models process the multiple input point clouds in different ways (no mixing, mixing + unmixing, mixing + masking), but are expected to output similar features, thereby compelling the network to extract essential features from the input. Extensive experiments show that our well-designed PatchMixing strategy and effective teacher-student learning architecture yield impressive results. Specifically, our model achieves a remarkable 92.9% classification accuracy in the Linear SVM task on the ModelNet40 dataset. Through pre-training and fine-tuning on downstream tasks, our method achieves an 89.8% classification accuracy on the most challenging split of ScanObjectNN and an outstanding 94.0% on ModelNet40.