Recent years have witnessed an increasing interest in image-based question-answering (QA) tasks. However, due to data limitations, there has been much less work on video-based QA. In this paper, we present TVQA, a large-scale video QA dataset based 6 popular TV shows. TVQA consists of 152,545 pairs from 21,793 clips, spanning over 460 hours video. Questions are designed be compositional nature, requiring systems jointly localize relevant moments within clip, comprehend subtitle-based...

The canonical approach to video-and-language learning (e.g., video question answering) dictates a neural model learn from offline-extracted dense features vision models and text language models. These feature extractors are trained independently usually on tasks different the target domains, rendering these fixed sub-optimal for downstream tasks. Moreover, due high computational overload of features, it is often difficult (or infeasible) plug directly into existing approaches easy...

We present the task of Spatio-Temporal Video Question Answering, which requires intelligent systems to simultaneously retrieve relevant moments and detect referenced visual concepts (people objects) answer natural language questions about videos. first augment TVQA dataset with 310.8K bounding boxes, linking depicted objects in answers. name this augmented version as TVQA+. then propose Answerer Grounded Evidence (STAGE), a unified framework that grounds evidence both spatial temporal...

Accurately and timely detecting multiscale small objects that contain tens of pixels from remote sensing images (RSI) remains challenging. Most the existing solutions primarily design complex deep neural networks to learn strong feature representations for separated background, which often results in a heavy computation burden. In this article, we propose an accurate yet fast object detection method RSI, named SuperYOLO, fuses multimodal data performs high-resolution (HR) on by utilizing...

Anomaly detection is one of the most important applications hyperspectral imaging technology. It a challenging task due to high dimensionality images (HSIs), redundant information, noisy bands, and limited capability utilizing spatial information. In this paper, we address these problems propose novel anomaly method in HSIs. Our approach, called structure tensor guided filter (STGF)-based strategy for detection, based on characteristics First, band selection algorithm proposed reduce...

Hyperspectral (HS) image can describe subtle differences in the spectral signatures of materials, but it has low spatial resolution limited by existing technical and budget constraints. In this paper, we propose a promising HS pansharpening method with deep priors (HPDP) to fuse low-resolution (LR) high-resolution (HR) panchromatic (PAN) image. Different from methods, redefine response function (SRF) based on larger eigenvalue structure tensor (ST) matrix for first time that is more line...

Reliable detection of anomalies without any prior information is a critical yet challenging task in many applications, not least military and civilian fields. An intelligent anomaly system would use the material-specific spectral hyperspectral images (HSIs), thereby avoiding loss visually confusing objects. However, conventional methods are mainly achieved an unsupervised way leading to limited performance due lack knowledge. In this article, we propose novel autoencoder adversarial-learning...

Hyperspectral images (HSIs) can describe the subtle differences in spectral signatures of materials. However, they have low spatial resolution due to various hardware limitations. Improving it via postprocess without an auxiliary high-resolution (HR) image still remains a challenging problem. In this paper, we address problem and propose new HSI super-resolution (SR) method. Our approach, called deep feature matrix factorization (DFMF), blends extracted by neural network (DNN) with...

In this article, we propose a weakly supervised low-rank representation (WSLRR) method for hyperspectral anomaly detection (HAD), which formulates deep learning-based HAD into low-lank optimization problem not only characterizing the complex and diverse background in real HSIs but also obtaining relatively strong supervision information. Different from existing unsupervised methods, first model manner, achieves better performance without prior information is restrained by richly correct...

Anomaly detection (AD) using hyperspectral images (HSIs) is of great interest for deep space exploration and Earth observations. This article proposes a weakly supervised discriminative learning with spectral constrained generative adversarial network (GAN) anomaly (HAD), called weaklyAD. It can enhance the discrimination between background homogenization saliency in cases where anomalous samples are limited sensitive to background. A novel probability-based category thresholding first...

Vision transformers (ViTs) have achieved impressive results on various computer vision tasks in the last several years. In this work, we study capability of frozen ViTs, pretrained only visual data, to generalize audio-visual data without finetuning any its original parameters. To do so, propose a latent hybrid (LAVISH) adapter that adapts ViTs by injecting small number trainable parameters into every layer ViT. efficiently fuse and audio cues, our LAVISH uses set tokens, which form an...

With recent advancements in aerospace technology, the volume of unlabeled remote sensing image (RSI) data has increased dramatically. Effectively leveraging this through self-supervised learning (SSL) is vital field sensing. However, current methodologies, particularly contrastive (CL), a leading SSL method, encounter specific challenges domain. Firstly, CL often mistakenly identifies geographically adjacent samples with similar semantic content as negative pairs, to confusion during model...

Hyperspectral anomaly detection faces various levels of difficulty due to the high dimensionality hyperspectral images (HSIs), redundant information, noisy bands, and limited capability utilizing spectral-spatial information. In this paper, we address these problems propose a novel approach, called feature extraction (SSFE), which is based on two main aspects. spectral domain, assume that anomalous pixels are rarely present all (or most) samples around anomalies belong background (BKG)....

Limited by the anomalous spectral vectors in unlabeled hyperspectral images (HSIs), anomaly detection methods based on background distribution estimation often suffer from contamination of anomalies, which decreases accuracy and, thus, weakens performance. To address this problem, we proposed a novel semisupervised learning (SSL) for framework generative adversarial network (GAN). GAN is applied and developed to estimate manner obtain an initial feature because its strong representational...

Hyperspectral (HS) pansharpening, as a special case of the superresolution (SR) problem, is to obtain high-resolution (HR) image from fusion an HR panchromatic (PAN) and low-resolution (LR) HS image. Though pansharpening based on deep learning has gained rapid development in recent years, it still challenging task because following requirements: 1) unique model with goal fusing two images different dimensions should enhance spatial resolution while preserving spectral information; 2) all...

Theoretically, hyperspectral images (HSIs) are capable of providing subtle spectral differences between different materials, but in fact, it is difficult to distinguish background and anomalies because the samples anomalous pixels HSIs limited susceptible noise. To explore discriminant features, a adversarial feature learning (SAFL) architecture specially designed for anomaly detection this article. In addition reconstruction loss, SAFL also introduces constraint loss network with batch...

Hyperspectral images (HSIs) have unique advantages in distinguishing subtle spectral differences of different materials. However, due to complex and diverse backgrounds, unknown prior knowledge, imbalanced samples, it is challenging separate background anomaly. In this article, we present a novel characterization background-anomaly separability with generative adversarial network (BASGAN) for hyperspectral anomaly detection. The key contribution the proposal explicitly constrain by...

The Smith-Waterman (S-W) algorithm is widely adopted by the state-of-the-art DNA sequence aligners. Existing wave front-based methods ignored fact that S-W fed with significantly varied-size inputs in modern aligners, which further optimized exerting extensive pruning. In this paper, we propose an architecture, tailored for varied input sizes as well harnessing software pruning strategies, to accelerate S-W. Our implementation demonstrates a 26.4x speedup over 24-thread Intel Has Xeon...

Burnout is a worldwide phenomenon among social welfare workers. This study examined how burnout affects student workers and professional It first the construct validity of Maslach Inventory‐General Survey (MBI‐GS) scale, using data from two Chinese samples (848 748 workers). The original three‐factor model was regarded as superior to other competing models. Investigation second‐order factor indicated that exhaustion cynicism are core components but personal efficacy not. research also...

Recently, autoencoder (AE)-based anomaly detection has drawn considerable interest in hyperspectral image (HSI) analysis. In this article, we propose a novel discriminative reconstruction method for images with spectral learning (SLDR). The proposed algorithm the following innovations. First, use error map (SEM) to detect anomalies because SEM can preferably reflect similarity of each pixel between input and reconstruction. Second, loss function SLDR model additionally introduces angle...