Long-tailed semi-supervised learning poses a significant challenge in training models with limited labeled data exhibiting a long-tailed label distribution. Current state-of-the-art LTSSL approaches heavily rely on high-quality pseudo-labels for large-scale unlabeled data. However, these methods often neglect the impact of representations learned by the neural network and struggle with real-world unlabeled data, which typically follows a different distribution than labeled data. This paper introduces a novel probabilistic framework that unifies various recent proposals in long-tail learning. Our framework derives the class-balanced contrastive loss through Gaussian kernel density estimation. We introduce a continuous contrastive learning method, CCL, extending our framework to unlabeled data using reliable and smoothed pseudo-labels. By progressively estimating the underlying label distribution and optimizing its alignment with model predictions, we tackle the diverse distribution of unlabeled data in real-world scenarios. Extensive experiments across multiple datasets with varying unlabeled data distributions demonstrate that CCL consistently outperforms prior state-of-the-art methods, achieving over 4% improvement on the ImageNet-127 dataset. Our source code is available at https://github.com/zhouzihao11/CCL
Ultrafast ultrasound imaging with plane wave transmission has been a promising technique to image moving objects, however, implies compromises among resolution, contrast and sensitivity. Coherence plane-wave compounding (CPWC) can balance the image quality and frame rate. The image quality, especiallyin terms of the suppression of artifacts stemmed from side lobes, is greatly comprised by reducing the number of the tilted plane-waves. However, in some special scenarios, such as tracking shear wave propagation inside soft tissue, and imaging the complex blood flow, it's better to keep a very high frame rate. How to realize a B-mode image of equivalent quality to the standard focused approach at a very high frame rate? Here we proposed a new imaging framework by combining CPWC with angle coherence factor. The B-mode images from simulation, experimental phantoms demonstrated that our proposed methodology greatly suppressed the side-lobes artifacts by around 20 dB compared with CPWC imaging, while the image quality, especially lateral resolution and contrast kept equivalent.
More and more attention has been given in the field of mechanical engineering to a material’s R-value, a parameter that characterizes the ability of sheet metal to resist thickness strain. Conventional methods used to determine R-value are based on experiments and an assumption of constant volume. Because the thickness strain cannot be directly measured, the R-value is currently determined using experimentally measured strains in the width, and loading directions in combination with the constant volume assumption, to determine the thickness strain indirectly. This paper provides an alternative method for determining the R-value without any assumptions. This method is based on the use of a multi-camera DIC system to measure strains in three directions simultaneously. Two sets of stereo-vision DIC measurement systems, each comprised of two GigE cameras, are placed on the front and back sides of the sample. Use of the double-sided calibration strategy unifies the world coordinate system of the front and back DIC measurement systems to one coordinate system, allowing for the measurement of thickness strain and direct calculation of R-value. The Random Sample Consensus (RANSAC) algorithm is used to eliminate noise in the thickness strain data, resulting in a more accurate R-value measurement.
This study aimed to identify the psychological variables that can mediate an influence of color focality on color preference in modern Japanese culture and to specify the pattern of the mediated influence. The potential mediators were 22 appraisal dimensions and the psychological processing fluency (PPF) of colors. We conducted two experiments with native Japanese speakers, which measured these variables. Data analyses showed that gracefulness is the only appraisal dimension that can mediate an effect of color focality on color preference. Specifically, focality negatively influenced the appraisal of gracefulness and gracefulness positively impacted preference. Although a U-shaped quadratic relationship exists between focality and PPF, no relationship was found between PPF and preference. Therefore, PPF's role as a mediator is not supported. Our literature survey suggests that the focality-gracefulness-preference relationship may be caused by the "noisy color phenomenon," which is a cultural phenomenon existing in many color-using areas in modern Japan.
Shearography is a coherent optical technique that allows the identification of the first derivative of deformation in the shearing direction. Due to direct measuring strain information, shearography is suited for non-destructive testing and evaluation (NDT/NDE). However, if there is a small defect parallel to the shearing direction, the first derivative of deformation in the direction has no noticeable change, and the defect is not visible. Therefore, the development of a shearography system with dual-directional simultaneous measurement of the first derivatives of deformation both in x- and y-directions is highly demanded in the field of NDT/NDE. It is suited to inspect complicated defects, such as long and narrow slots, microcracks, etc. This paper presents a review of shearography for different dual-directional systems developed in the last two decades. After a brief overview of shearography, the paper will display two dual-directional shearographic techniques—temporal phase-shift (TPS) and spatial phase-shift (SPS) methods. TPS dual-shearing systems are suited for static measurements, while the SPS dual-shearing systems are useful for dynamic measurements. The basic theories, optical layouts, and comparisons are presented. The advantages and disadvantages of practical applications are discussed.
Past studies have reported that language-specific color focality has substantial influence on short-term memory (STM) performance of colors of the speakers of the language, which we call the “focality effect.” This study attempts to clarify the continuous patterns of this effect, that is, the manner in which correct recognition possibilities and misrecognition error distances of colors, which are two aspects of the STM performance of colors, change in a gradual fashion along the continuum of color focality. Our experiment, which tests the Japanese language, finds that a U-shaped relationship exists between the focality and the possibility of correct recognition, and that the misrecognition error distance increases as the focality decreases. We speculate that the subjects' frequent and conscious employment of the memorization strategy of coding colors using linguistic color categories is one important cause of the detected effect patterns.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)