Search

Recent developments in the field of non-local attention (NLA) have led to a renewed interest in self-similarity-based single image super-resolution (SISR). Researchers usually use the NLA to explore non-local self-similarity (NSS) in SISR and achieve satisfactory reconstruction results. However, a surprising phenomenon that the reconstruction performance of the standard NLA is similar to that of the NLA with randomly selected regions prompted us to revisit NLA. In this paper, we first analyzed the attention map of the standard NLA from different perspectives and discovered that the resulting probability distribution always has full support for every local feature, which implies a statistical waste of assigning values to irrelevant non-local features, especially for SISR which needs to model long-range dependence with a large number of redundant non-local features. Based on these findings, we introduced a concise yet effective soft thresholding operation to obtain high-similarity-pass attention (HSPA), which is beneficial for generating a more compact and interpretable distribution. Furthermore, we derived some key properties of the soft thresholding operation that enable training our HSPA in an end-to-end manner. The HSPA can be integrated into existing deep SISR models as an efficient general building block. In addition, to demonstrate the effectiveness of the HSPA, we constructed a deep high-similarity-pass attention network (HSPAN) by integrating a few HSPAs in a simple backbone. Extensive experimental results demonstrate that HSPAN outperforms state-of-the-art approaches on both quantitative and qualitative evaluations. Our code and a pre-trained model were uploaded to GitHub (https://github.com/laoyangui/HSPAN) for validation. [ABSTRACT FROM AUTHOR]

Contrastive self-supervised representation learning methods maximize the similarity between the positive pairs, and at the same time tend to minimize the similarity between the negative pairs. However, in general the interplay between the negative pairs is ignored as they do not put in place special mechanisms to treat negative pairs differently according to their specific differences and similarities. In this paper, we present Extended Momentum Contrast (XMoCo), a self-supervised representation learning method founded upon the legacy of the momentum-encoder unit proposed in the MoCo family configurations. To this end, we introduce a cross consistency regularization loss, with which we extend the transformation consistency to dissimilar images (negative pairs). Under the cross consistency regularization rule, we argue that semantic representations associated with any pair of images (positive or negative) should preserve their cross-similarity under pretext transformations. Moreover, we further regularize the training loss by enforcing a uniform distribution of similarity over the negative pairs across a batch. The proposed regularization can easily be added to existing self-supervised learning algorithms in a plug-and-play fashion. Empirically, we report a competitive performance on the standard Imagenet-1K linear head classification benchmark. In addition, by transferring the learned representations to common downstream tasks, we show that using XMoCo with the prevalently utilized augmentations can lead to improvements in the performance of such tasks. We hope the findings of this paper serve as a motivation for researchers to take into consideration the important interplay among the negative examples in self-supervised learning. [ABSTRACT FROM AUTHOR]

Existing person re-identification (Re-ID) methods usually rely heavily on large-scale thoroughly annotated training data. However, label noise is unavoidable due to inaccurate person detection results or annotation errors in real scenes. It is extremely challenging to learn a robust Re-ID model with label noise since each identity has very limited annotated training samples. To avoid fitting to the noisy labels, we propose to learn a prefatory model using a large learning rate at the early stage with a self-label refining strategy, in which the labels and network are jointly optimized. To further enhance the robustness, we introduce an online co-refining (CORE) framework with dynamic mutual learning, where networks and label predictions are online optimized collaboratively by distilling the knowledge from other peer networks. Moreover, it also reduces the negative impact of noisy labels using a favorable selective consistency strategy. CORE has two primary advantages: it is robust to different noise types and unknown noise ratios; it can be easily trained without much additional effort on the architecture design. Extensive experiments on Re-ID and image classification demonstrate that CORE outperforms its counterparts by a large margin under both practical and simulated noise settings. Notably, it also improves the state-of-the-art unsupervised Re-ID performance under standard settings. Code is available at https://github.com/mangye16/ReID-Label-Noise. [ABSTRACT FROM AUTHOR]

Our impression about one person often updates after we see more aspects of him/her and this process keeps iterating given more meetings. We formulate such an intuition into the problem of person re-identification (re-ID), where the representation of a query (probe) image is iteratively updated with new information from the candidates in the gallery. Specifically, we propose a simple attentional aggregation formulation to instantiate this idea and showcase that such a pipeline achieves competitive performance on standard benchmarks including CUHK03, Market-1501 and DukeMTMC. Not only does such a simple method improve the performance of the baseline models, it also achieves comparable performance with latest advanced re-ranking methods. Another advantage of this proposal is its flexibility to incorporate different representations and similarity metrics. By utilizing stronger representations and metrics, we further demonstrate state-of-the-art person re-ID performance, which also validates the general applicability of the proposed method. [ABSTRACT FROM AUTHOR]

With the increasing popularity of mobile phones and tablets, the explosive growth of query-by-capture applications calls for a compact representation of the query image feature. Compact descriptors for visual search (CDVS) is a recently released standard from the ISO/IEC moving pictures experts group, which achieves state-of-the-art performance in the context of image retrieval applications. However, they did not consider the matching characteristics in local space in a large-scale database, which might deteriorate the performance. In this paper, we propose a more compact representation with scale invariant feature transform (SIFT) descriptors for the visual query based on Grassmann manifold. Due to the drastic variations in image content, it is not sufficient to capture all the information using a single transform. To achieve more efficient representations, a SIFT manifold partition tree (SMPT) is initially constructed to divide the large dataset into small groups at multiple scales, which aims at capturing more discriminative information. Grassmann manifold is then applied to prune the SMPT and search for the most distinctive transforms. The experimental results demonstrate that the proposed framework achieves state-of-the-art performance on the standard benchmark CDVS dataset. [ABSTRACT FROM AUTHOR]

Image descriptors based on activations of Convolutional Neural Networks (CNNs) have become dominant in image retrieval due to their discriminative power, compactness of representation, and search efficiency. Training of CNNs, either from scratch or fine-tuning, requires a large amount of annotated data, where a high quality of annotation is often crucial. In this work, we propose to fine-tune CNNs for image retrieval on a large collection of unordered images in a fully automated manner. Reconstructed 3D models obtained by the state-of-the-art retrieval and structure-from-motion methods guide the selection of the training data. We show that both hard-positive and hard-negative examples, selected by exploiting the geometry and the camera positions available from the 3D models, enhance the performance of particular-object retrieval. CNN descriptor whitening discriminatively learned from the same training data outperforms commonly used PCA whitening. We propose a novel trainable Generalized-Mean (GeM) pooling layer that generalizes max and average pooling and show that it boosts retrieval performance. Applying the proposed method to the VGG network achieves state-of-the-art performance on the standard benchmarks: Oxford Buildings, Paris, and Holidays datasets. [ABSTRACT FROM AUTHOR]

Accurate distance perception and natural interactions are mandatory conditions when training precision aiming tasks in VR. However, many factors specific to virtual environments (VEs) lead to differences in the way users execute a motor task in VR versus the real world. To investigate these differences, the authors performed a study on basketball beginners' free-throw performance in VEs under different visual conditions. Although the success rate is not statistically different, some adaptations occurred in the way the users performed the task, depending on the visual conditions. In the third-person perspective visual condition, the release parameters indicate that the users more accurately estimated distance to target. Adding visual guidance information (gradual depth information showing the ideal ball trajectory) also led to more natural motor behavior. The final aim of this study was to develop a reliable basketball free-throw training system in VEs, so the authors compared beginners' performances in VR with experts' models of performance. Their results show that most of the performance variables tended to evolve closer to the experts' performance during the training in the VE. [ABSTRACT FROM PUBLISHER]