Candidate Images Research Articles

Integrating listwise ranking into pairwise-based image-text retrieval

Image-Text Retrieval (ITR) is essentially a ranking problem. Given a query caption, the goal is to rank candidate images by relevance, from large to small. The current ITR datasets are constructed in a pairwise manner. Image-text pairs are annotated as positive or negative. Correspondingly, ITR models mainly use pairwise losses, such as triplet loss, to learn to rank. Pairwise-based ITR increases positive pair similarity while decreasing negative pair similarity indiscriminately. However, the relevance between dissimilar negative pairs is different. Pairwise annotations cannot reflect this difference in relevance. In the current datasets, pairwise annotations miss many correlations. There are many potential positive pairs among the pairs labeled as negative. Pairwise-based ITR can only rank positive samples before negative samples, but cannot rank negative samples by relevance. In this paper, we integrate listwise ranking into conventional pairwise-based ITR. Listwise ranking optimizes the entire ranking list based on relevance scores. Specifically, we first propose a Relevance Score Calculation (RSC) module to calculate the relevance score of the entire ranked list. Then we choose the ranking metric, Normalised Discounted Cumulative Gain (NDCG), as the optimization objective. We apply a metric smoothing method named Smooth-NDCG (S-NDCG) to ITR, which transforms the non-differentiable NDCG into a differentiable listwise loss. Our listwise ranking approach can be plug-and-play integrated into current pairwise-based ITR models. Experiments on ITR benchmarks show that integrating listwise ranking can improve the performance of current ITR models and provide more user-friendly retrieval results. The code is available at https://github.com/AAA-Zheng/Listwise_ITR.

Stochastic Reconstruction of 3D Heterogeneous Microstructure Using a Column-Oriented Multiple-Point Statistics Program

Abstract Three-dimensional (3D) microstructure reconstruction is a key approach to exploring the relationship between pore characteristics and physical properties. Viewing the training image as a prior model, multiple-point statistics (MPS) focus on reproducing spatial patterns in the simulation grid. However, it is challenging to efficiently generate 3D nonstationary models with varying microstructures. In this work, we propose column-oriented simulation (ColSIM) to achieve the stochastic reconstruction of 3D porous media. A heterogeneous system is understood as a spatially evolving process that consists of frequent transitions of small magnitude and abrupt changes of large magnitude. First, a training image selection step is suggested to find representative microstructures. Our program applies modified Hausdorff distance, t-distributed stochastic neighboring embedding, and spectral clustering to organize two-dimensional (2D) candidate images. The medoid of each group is applied to guide the following programs. Second, we introduce column-oriented searching into MPS. To save simulation time, a subset of conditioning points is checked to find desired instances. Our program suggests an early stopping strategy to address complex microstructures. Third, a contrastive loss term is designed to create 3D models from 2D slice. To automatically calibrate the volume fraction and simplify parameter specification, the computer consistently monitors the difference between the present model and the target. The performance of ColSIM is examined by 3D multiphase material modeling and 3D heterogeneous shale simulation. To achieve quantitative evaluation, we compute various statistical functions and physical descriptors on simulated realizations. The proposed ColSIM exhibits competitive performance in terms of calculation efficiency, microstructure reproduction, and spatial uncertainty.

A hierarchical probabilistic underwater image enhancement model with reinforcement tuning

Underwater Image Enhancement (UIE) is a challenging problem due to the complex underwater environment. Traditional UIE methods can hardly adapt to various underwater environments. Deep learning-based UIE methods are more powerful but often rely on a large deal of real-world underwater images with distortion-free reference images. This gives rise to two issues: First, the reference images are highly uncertain because the ground-truth images cannot be are captured directly in underwater environment. Second, learning-based methods may lack generalization ability for diverse underwater environments. To tackle these issues, we propose HPUIE-RL, a hierarchical probabilistic UIE model facilitated by reinforcement learning. This model integrates UNet with hierarchical probabilistic modules to produce various enhanced candidate images that reflect the uncertainty of the enhancement. Then, a reinforcement learning fine-tuning framework is designed to fine-tune the pretrained model in an unsupervised manner, which responds to the dynamic underwater environment. Experiments on real-world datasets from diverse underwater environments demonstrate that our HPUIE-RL model outperforms state-of-the-art UIE methods regarding visual and quantitative performance and generalizability.

The Mediating Impact of Candidate Image on the Relationship between Religiosity and Voting Intention

This study aims to identify the influence of reliability on voting intentions. The research also investigates the role of the candidate image variable as a meditation between religiosity and voting intention. Four hypotheses were developed and tested using a quantitative approach. The theory used in this study is the theory of reasoned action. Primary research data was collected using the questionnaire through purposive sampling. The variable is measured using a Likert scale of five with a range of “very disagree (1) to "very agree." The data processing method applies the structural equation model using Smart-Partial Least Square (PLS) 3.0 software. The final sample consists of 155 respondents. Measurement model evaluation and structural model assessment are used to obtain research results. After passing the assessment of the measurement model, the structural model shows that the model has good relevance and predictive strength. The hypothesis testing indicates that candidate image mediates the relationship between religiosity and voting intention, and candidate images mediate the relation between religiosity and voting intention with a mediating complementary mediation. Furthermore, the results also show that four direct influences are also significant: (i) religiosity on candidate image (ii) candidate image on voting intention; (iii) candidate image mediates relations between the influence of religiosity and voter intention.

Research on Pedestrian Re-identification Techniques in Dynamic Scenes Using Convolutional Neural Networks

Abstract Pedestrian re-recognition is the process of retrieving pedestrians with the same identity information as a given pedestrian from a cross-domain view candidate image dataset or a non-overlapping surveillance video sequence using computer vision techniques. The goal of this paper is to use convolutional neural network techniques to re-recognize pedestrians in dynamic scenes. Through the use of convolutional calculations, activation function selection, and other techniques, this paper provides basic technical support for the research of pedestrian re-recognition technology. A Siamese network is obtained by applying convolutional neural networks to pedestrian recognition as the main discriminative model for subsequent research. In order to effectively solve the problem of occlusion, illumination, and other possible interference with the recognition effect in dynamic scenes, this paper adopts the image enhancement method of random erasure and introduces the attention mechanism to improve the robustness of the model to the occlusion of pedestrian images. Through the examination of the model on the dataset of the average accuracy mean (MAP) and other indicators and the actual application in the construction site and other scenes, it is proved that the pedestrian re-recognition model used in this paper has a more significant recognition performance compared with other models, and can still maintain more than 80% of the accuracy rate under the application of dynamic and complex scenes.

Post-Mortem Analysis of the 2016 Elections Based on the Political Marketing Strategies Used by Local Political Parties in Dumaguete City

This research study is a post-mortem analysis of the 2016 local elections in Dumaguete City, focusing on the political marketing strategies used by the Liberal Party (LP) and Nationalist People’s Coalition (NPC) candidates. The objective was to analyze the political marketing strategies the political parties and candidates employed in the 2016 elections. It also aimed to address the research gap by systematically examining and analyzing the political marketing practices within the local context. The researcher employed a combination of descriptive-quantitative and qualitative designs. The quantitative design was facilitated by measuring the level of effectiveness of the political marketing strategies used by the party and the candidates to win votes, which was measured on a scale of 5.00, where 1.00 is not effective, and 5.00 is the most effective. The qualitative design was facilitated by the archival of documents and structured interviews. The study's respondents were the mayor, vice mayor, and ten city councilors, totaling 12 (n=12). However, only 9 (n=9) participated in answering the survey questionnaire and guide interview. Using Menon’s framework, the study found that LP dominated the election, employing strategies like clear platforms, candidate image building, party disclosure, information drives, material rewards, smart slogans, and social media use. They also managed attitudinal and behavioral barriers effectively. LP’s dominant strategies included media campaigns, house-to-house campaigns, public forums, small group meetings, and leaflet distribution. NPC, on the other hand, focused on good governance, transparency, and monetary indulgence. Both parties utilized product function, distribution function, cost function, machine politics and rewards, news management function, and communication function. The top three effective strategies were the news management function (4.61), product function (4.59), and communication function (4.58). The study concluded that LP was more thorough and organized in its marketing strategy application, giving them critical advantages in the election.

Modeling of intelligent hyperparameter tuned deep learning based copy move image forgery detection technique

Due to the drastic increase in the generation of high-quality fake images in social networking, it is essential to design effective recognition approaches. Image/video manipulation defines any set of actions which can be carried out on digital content by the use of software editing approaches or artificial intelligence. A major kind of image and video editing comprises replicating the regions of the image, named as copy-move technique. Conventional image processing methods physically search for the pattern relevant to the replicated contents, restricting the utilization in massive classification of data. Contrastingly, the recently developed deep learning (DL) models have exhibited promising performance over the traditional models. In this aspect, this paper presents a novel intelligent deep learning based copy move image forgery detection (IDL-CMIFD) technique. The proposed IDL-CMIFD technique intends to design a DL model to classify the candidate images into two classes: original and forged/tampered and then localized the copy moved regions. In addition, the proposed IDL-CMIFD technique involves the Adam optimizer with Efficient Net based feature extractor to derive a useful set of feature vectors. Moreover, chaotic monarch butterfly optimization (CMBO) with deep wavelet neural network (DWNN) model is applied for classification purposes. The CMBO algorithm is utilized to optimally tune the parameters involved in the DWNN model in such a way that the classification performance gets improved. The performance validation of the proposed model takes place on benchmark MICC-F220, MICC-F2000, MICC-F600 datasets. A wide range of comparative analyses is performed and the results ensured the better performance of the IDL-CMIFD technique in terms of different evaluation parameters.

Pengaruh Iklan Politik Partai Dan Citra Kandidat Terhadap Sikap Memilih

In every general election, various party political advertisements appear to persuade voters to give their support. In addition, the candidate's image is made as good as possible to meet the tastes of voters so that voters provide support to the candidate. This research was conducted to determine the effect of party political advertising and candidate image on voting attitudes of people in the city of Gorontalo. This study used a quantitative approach with explanatory survey methods. Data was collected using a questionnaire. The sample in this study were 383 residents of the city of Gorontalo and were taken using the Proportional Stratified Random Sampling technique. Data analysis in this study used Partial Least Square (PLS) with the help of the SmartPLS version 4 software program. The findings of this study indicate that there is no effect of party political advertising on voting attitudes of the people of Gorontalo city, candidate image has a significant effect on the voting attitude of the people of Gorontalo city and there is a difference in the effect of party political advertising and candidate image on voting attitudes of the people of Gorontalo city.

Research on the Two-way Interaction Between Voters and Political Parties in Elections

In democratic politics in the Western world, the process of elections are full with direct interactions with voters. Hence, voters choice affects whether a political party can become the ruling party. Voters prefer to choose parties that are in line with their own interests, and political parties need voters support to win the election. This paper studies the two-way interaction between voters and political parties in elections. The needs of target voters will affect the political views and positions of political parties. Therefore, in order to gain support, the party will promise to carry out some policies that would benefit the target voters. While the voters affect political parties positions, the parties also affect the voters choices, and Brexit is an example worth studying. Through media and advertisements, parties promote their core values and achievements, shape the image of candidates, and belittle the opponents. In this way, the parties can affect the votes preferable party and thus win the election.

Image, issue, and affective priming effects of candidates’ Facebook posts on voters’ trait perceptions and overall evaluations

ABSTRACT Trait perception is crucial to electoral support. Increasingly, candidates are using social media posts to shape their images, prime trait perception, and enhance overall evaluations. This study distinguishes three types of priming effects – namely, image priming, issue priming, and affective priming. Three online experiments assess these effects in the context of Facebook campaigns. The image priming effects were consistently confirmed. When coupled with issue discussions, their effectiveness was amplified only for applicable issues in positive tones. The suitability of applying media priming theory to social media campaigns is also discussed.

CLOOME: contrastive learning unlocks bioimaging databases for queries with chemical structures.

The field of bioimage analysis is currently impacted by a profound transformation, driven by the advancements in imaging technologies and artificial intelligence. The emergence of multi-modal AI systems could allow extracting and utilizing knowledge from bioimaging databases based on information from other data modalities. We leverage the multi-modal contrastive learning paradigm, which enables the embedding of both bioimages and chemical structures into a unified space by means of bioimage and molecular structure encoders. This common embedding space unlocks the possibility of querying bioimaging databases with chemical structures that induce different phenotypic effects. Concretely, in this work we show that a retrieval system based on multi-modal contrastive learning is capable of identifying the correct bioimage corresponding to a given chemical structure from a database of ~2000 candidate images with a top-1 accuracy >70 times higher than a random baseline. Additionally, the bioimage encoder demonstrates remarkable transferability to various further prediction tasks within the domain of drug discovery, such as activity prediction, molecule classification, and mechanism of action identification. Thus, our approach not only addresses the current limitations of bioimaging databases but also paves the way towards foundation models for microscopy images.

Boosting Diversity in Visual Search with Pareto Non-Dominated Re-Ranking

The field of visual search has gained significant attention recently, particularly in the context of web search engines and e-commerce product search platforms. However, the abundance of web images presents a challenge for modern image retrieval systems, as they need to find both relevant and diverse images that maximize users’ satisfaction. In response to this challenge, we propose a non-dominated visual diversity re-ranking (NDVDR) method based on the concept of Pareto optimality. To begin with, we employ a fast binary hashing method as a coarse-grained retrieval procedure. This allows us to efficiently obtain a subset of candidate images for subsequent re-ranking. Fed with this initial retrieved image results, the NDVDR performs a fine-grained re-ranking procedure for boosting both relevance and visual diversity among the top-ranked images. Recognizing the inherent conflict nature between the objectives of relevance and diversity, the re-ranking procedure is simulated as the analytical stage of a multi-criteria decision-making process, seeking the optimal tradeoff between the two conflicting objectives within the initial retrieved images. In particular, a non-dominated sorting mechanism is devised that produces Pareto non-dominated hierarchies among images based on the Pareto dominance relation. Additionally, two novel measures are introduced for the effective characterization of the relevance and diversity scores among different images. We conduct experiments on three popular real-world image datasets and compare our re-ranking method with several state-of-the-art image search re-ranking methods. The experimental results validate that our re-ranking approach guarantees retrieval accuracy while simultaneously boosting diversity among the top-ranked images.

Diversified Cover Selection for Image Steganography

This paper proposes a new cover selection method for steganography. We focus on the scenario that the available images for selection contain diversified sources, i.e., nature images and metaverse images. For the scenario, we design a targeted strategy to evaluate the suitability for steganography of a candidate image, which selects images according to the undetectability against steganalytic tools symmetrically. Firstly, steganalytic features of the candidate images are extracted. Then, the features are fed on a steganalytic classifier, and the possibility of carrying secret data is calculated for cover selection. As a result, the selected images are the best candidates to resist steganalysis. Experimental results show that our method performs better than existing cover selection schemes when checked by steganalytic tools.

PS J2107−1611: A new wide-separation, quadruply imaged lensed quasar with flux ratio anomalies

We report the discovery of PS J2107−1611, a fold-configuration 4.3″-separation quadruply lensed quasar with a bright lensed arc. It was discovered using a convolutional neural network on Pan-STARRS gri images of pre-selected quasar candidates with multiple nearby Pan-STARRS detections. Spectroscopic follow-up with EFOSC2 on the ESO 3.58 m New Technology Telescope reveals the source to be a quasar at z = 2.673, with the blended fold image pair showing deformed broad lines relative to the other images. The flux ratios measured from optical to near-infrared imaging in the Canada-France-Hawaii Telescope Legacy Survey, Pan-STARRS, the Legacy Surveys, and the Vista Hemisphere Survey are inconsistent with a smooth mass model as the fold pair images are ∼15 times too faint. Variability, time delay effects, and reddening are ruled out through multiple-epoch imaging and color information. The system is marginally resolved in the radio in the Very Large Array Sky Survey S-band, where it has a 10 mJy detection. The radio flux ratios are compatible with the smooth mass macromodel. This system offers a unique tool for future studies of quasar structure with strong and microlensing. A more detailed analysis of follow-up with JWST/MIRI, VLT/MUSE, VLT/ERIS, and data from the European Very Long Baseline Interferometer will be presented in a forthcoming paper.

Politics as a Spectacle: The Role of Advertising and Physical Image in Visualizing a Political Candidate as Merchandise and Their Impact on Voting Intentions

Advertising no longer describes objects but can impose on Society the obligation to consume whatever is produced. In this way, political advertising has become more commercial, bringing new ways of doing politics where advertising strategies achieve an adapted policy. New advertising strategies have turned the political process into a spectacle encouraging new political behaviour. Through the Society of the Spectacle and the Visual Frame, this quantitative study analysed how advertising uses the physical image of a political candidate as the merchandise that voters acquire as a force of social relationships. In addition, this study is novel in exploring the role of political ideology as a moderating variable that may dampen the effects between image advertising and voting intention. With 582 participants in an electronic survey analysed through PLS-SEM, the study reflects that, during the spectacle process, the political candidate can generate relationships, value and satisfaction through their physical attributes. It enables a purchase through the vote since, once the perception is manifested, the frame is accepted, leading the voter to make non-objective decisions. This research contributes new knowledge to the Spectacle Society and Visual Frame theory by providing a new way of analyzing political candidates and how advertising activates political behaviour.

How to convince in a televised debate: the application of machine learning to analyze why viewers changed their winner perception during the 2021 German chancellor discussion

What causes viewers to change their winner perception during a televised debate? The article addresses this question, drawing on a large-N field study of the 2021 chancellor debate in Germany, which contains survey and real-time response data for 4613 participants. Using machine learning techniques, we identify determinants of why participants change their opinion about who is winning the discussion during the debate. Our analysis based on random forest and decision tree models shows in detail, first, what factors drive debate winner perceptions in the course of televised debate reception. Second, we reveal what combinations of political predispositions and candidate statements are necessary to change the viewers’ debate winner perception. In doing so, third, we expand the toolbox of empirical debate research with our analysis based on machine learning algorithms. Our findings indicate that pre-debate chancellor preference and candidate images play a crucial role in determining post-debate perception change, while party identification is less important in predicting changes. We can also directly identify several speech moments in the debate that shifted viewer perception, a novel approach to evaluating political debate performance.

Enhancing Lane-Tracking Performance in Challenging Driving Environments through Parameter Optimization and a Restriction System

The autonomous driving market has experienced rapid growth in recent times. From systems that assist drivers in keeping within their lanes to systems that recognize obstacles using sensors and then handle those obstacles, there are various types of systems in autonomous driving. The sensors used in autonomous driving systems include infrared detection devices, lidar, ultrasonic sensors, and cameras. Among these sensors, cameras are widely used. This paper proposes a method for stable lane detection from images captured by camera sensors in diverse environments. First, the system utilizes a bilateral filter and multiscale retinex (MSR) with experimentally optimized set parameters to suppress image noise while increasing contrast. Subsequently, the Canny edge detector is employed to detect the edges of the lane candidates, followed by utilizing the Hough transform to make straight lines from the land candidate images. Then, using a proposed restriction system, only the two lines that the current vehicle is actively driving within are detected from the candidate lines. Furthermore, the lane position information from the previous frame is combined with the lane information from the current frame to correct the current lane position. The Kalman filter is then used to predict the lane position in the next frame. The proposed lane-detection method was evaluated in various scenarios, including rainy conditions, low-light nighttime environments with minimal street lighting, scenarios with interfering guidelines within the lane area, and scenarios with significant noise caused by water droplets on the camera. Both qualitative and quantitative experimental results demonstrate that the lane-detection method presented in this paper effectively suppresses noise and accurately detects the two active lanes during driving.

Using photovoice to explore attitudes toward and experiences with images of older adults

ABSTRACT This study examined university students’ perceptions of older adults. The study employed the photovoice, analyzing visual data (photographs) and in-depth interviews conducted with five university students to capture more accurate and candid images of older adults. Themes from photographs, interviews, and focus group discussions were organized into three categories: study participants’ negative ‘otherization’ of older adults, a critical awakening of ‘discrimination and exclusion’ in daily life, and the consciousness raising among young people who were used to ‘otherizing’. Young adults who differed from older adults due to their decline in old age noticed various forms and degrees of discrimination and exclusion of older adults in daily life through photovoice experience. Moreover, young adults developed positive perspectives and recognized the strengths of the older adults. This study is unique in providing specific and practical insights into the co-existence of two generations. Our findings suggest that developing and implementing programs and policies that promote positive interaction between young and older generations will contribute to resolving inter-generational conflict and ageism.

Candidate Image and Voting Intention in First Time Voters: Meta-Analysis Study

Candidate Image and Voting Intention in First Time Voters: Meta-Analysis Study

Scene-Level Sketch-Based Image Retrieval with Minimal Pairwise Supervision

The sketch-based image retrieval (SBIR) task has long been researched at the instance level, where both query sketches and candidate images are assumed to contain only one dominant object. This strong assumption constrains its application, especially with the increasingly popular intelligent terminals and human-computer interaction technology. In this work, a more general scene-level SBIR task is explored, where sketches and images can both contain multiple object instances. The new general task is extremely challenging due to several factors: (i) scene-level SBIR inherently shares sketch-specific difficulties with instance-level SBIR (e.g., sparsity, abstractness, and diversity), (ii) the cross-modal similarity is measured between two partially aligned domains (i.e., not all objects in images are drawn in scene sketches), and (iii) besides instance-level visual similarity, a more complex multi-dimensional scene-level feature matching problem is imposed (including appearance, semantics, layout, etc.). Addressing these challenges, a novel Conditional Graph Autoencoder model is proposed to deal with scene-level sketch-images retrieval. More importantly, the model can be trained with only pairwise supervision, which distinguishes our study from others in that elaborate instance-level annotations (for example, bounding boxes) are no longer required. Extensive experiments confirm the ability of our model to robustly retrieve multiple related objects at the scene level and exhibit superior performance beyond strong competitors.