Visual Attributes Research Articles

Parallel processing of value-related information during multi-attribute decisions.

When choosing between options with multiple attributes, do we integrate all attributes into a unified measure for comparison, or does the comparison also occur at the level of each attribute, involving parallel processes that can dynamically influence each other? What happens when independent sensory features all carry information about the same decision factor, such as reward value? To investigate these questions, we asked human participants to perform a two-alternative forced choice reaching task in which the reward value of a target was indicated by two visual attributes-its brightness ("bottom-up," BU feature) and its orientation ("top-down," TD feature). If decisions always occur after the integration of both features, there should be no difference in the reaction time (RT) regardless of the attribute combinations that drove the choice. Counter to that prediction, RT distributions depended on the attribute combinations of given targets and the choices made by participants. RTs were shortest when both attributes were congruent or when the choice was based on the bottom-up feature, and longer when the attributes were in conflict (favoring opposite options). In conflict trials, nearly two-thirds of participants made faster decisions when choosing the option favored by the bottom-up feature than when choosing the top-down-favored option. We also observed mid-reach changes-of-mind in a subset of conflict trials, mostly changing from the bottom-up to the top-down-favored target. These data suggest that multi-attribute value-based decisions are better explained by a distributed process including competition among different features than by a competition based on a single, integrated estimate of value.NEW & NOTEWORTHY We show that during value-based decisions, humans do not always use all reward-related information to make their choice, but instead can "jump the gun" using partial information. In particular, when different sources of information were in conflict, early decisions were mostly based on fast bottom-up information, and sometimes followed by corrective changes-of-mind based on slower top-down information. This supports parallel decision processes among different information sources, as opposed to a single integrated "common currency."

A Multimodal User-Adaptive Recommender System

Traditional recommendation systems have predominantly relied on user-provided ratings as explicit input. Concurrently, visually aware recommender systems harness inherent visual cues within data to decode item characteristics and deduce user preferences. However, the untapped potential of incorporating item images into the recommendation process warrants investigation. This paper introduces an original convolutional neural network (CNN) architecture that leverages multimodal information, connecting user ratings with product images to enhance item recommendations. A central innovation of the proposed model is the User-Adaptive Filtering Module, a dynamic component that utilizes user profiles to generate personalized filters. Through meticulous visual influence analysis, the effectiveness of these filters is demonstrated. Furthermore, experimental results underscore the competitive performance of the approach compared to traditional collaborative filtering methods, thereby offering a promising avenue for personalized recommendations. This approach capitalizes on user adaptation patterns, enhancing the understanding of user preferences and visual attributes.

Theme-Aware Visual Attribute Reasoning for Image Aesthetics Assessment

People usually assess image aesthetics according to visual attributes, e.g., interesting content, good lighting and vivid color, etc. Further, the perception of visual attributes depends on the image theme. Therefore, the inherent relationship between visual attributes and image theme is crucial for image aesthetics assessment (IAA), which has not been comprehensively investigated. With this motivation, this paper presents a new IAA model based on Theme-Aware Visual Attribute Reasoning (TAVAR). The underlying idea is to simulate the process of human perception in image aesthetics by performing bilevel reasoning. Specifically, a visual attribute analysis network and a theme understanding network are first pre-trained to extract aesthetic attribute features and theme features, respectively. Then, the first level Attribute-Theme Graph (ATG) is built to investigate the coupling relationship between visual attributes and image theme. Further, a flexible aesthetics network is introduced to extract general aesthetic features, based on which we built the second level Attribute-Aesthetics Graph (AAG) to mine the relationship between theme-aware visual attributes and aesthetic features, producing the final aesthetic prediction. Extensive experiments on four public IAA databases demonstrate the superiority of the proposed TAVAR model over the state-of-the-arts. Furthermore, TAVAR features better explainability due to the use of visual attributes.

Impact of Surgical Rejuvenation on Visual Processing and Character Attribution of Faces.

This study considers observers' reflexive responses to the rejuvenated face, and how instinctive responses relate to subjective judgment. We investigated observers' reflexive perception of faces both pre and post surgical intervention during the early stages of visual processing. Subjective character attribution for all test images was also assessed by the same observers. Forty frontal facial images of 20 patients portraying the pre- and postoperative high superficial musculoaponeurotic system facelift along with variable concomitant procedures were studied. Nineteen lookzone regions were mapped post hoc onto each image. Forty observers examined the images, whereas an eye-tracking camera recorded their eye movements. Visual fixation data were recorded and analyzed. Observers also rated each image on the basis of five elemental positive character attributes. A statistically coherent but nonsignificant (P > 0.05) trend was identified with the surgical intervention resulting in greater attention being paid to the central triangle region of the face with reduction in attention to the facial periphery. Facial rejuvenation significantly increased the subjective character ratings of all five positively valenced attributes tested. Average age estimate of the photos decreased significantly from 54 to 48.6 years (true average age of 57.4 years). We provide data illustrating both reflexive and subjective responses to facial rejuvenation. Observers reported a more favorable impression of the treated faces and evaluated them as being younger than their true age. A trend was detected for increased visual fixation of the central facial region following rejuvenation. Interpretation of these findings and indication for further research is provided.

A Computer Vision Methodology to Predict Brand Personality from Image Features

Using the computer vision method, this study proposes an analytical model of visual aesthetics for brand communication and analyzes the effects of visual features (i.e., colors and visual complexity) on brand personality. This study illustrates a four-step procedure correlating computationally coded visual attributes with human ratings of perceived brand personality. This study has important methodological implications for advertising researchers and practitioners.

Digital fabrics for online shopping and fashion design

Improving the digital presentation of fabrics enhances the online shopping experience and, in turn, reduces textile waste. In this study, we examined how the manipulation of simple surface reflectance models can bias the perception of fabric properties simulated online in a web browser. We showed that motion and three-dimensional (3D) folds (i.e., rumple) influence the perception of sheen for different fabric types (cotton knit and satin). Also, we found complex interactions between these parameters in their effects on perceived sheen and perceived color saturation. Moreover, we showed that changing the level of specular roughness significantly influences visual perception of sheen, color and lightness, which in turn, can categorically alter perceptual judgments of material type. In contrast to visual attributes, specular roughness did not influence visually perceived tactile characteristics of digital fabrics (thickness and stretch). The knowledge gained about perceptual biases of digital fabrics from this study will inform future considerations for optimizing the fidelity of textiles depicted in digital commerce.

Spatio-Temporal Experience of Tour Routes in the Humble Administrator’s Garden Based on Isovist Analysis

Chinese classical gardens (CCGs), as a distinct spatial category within architectural, historical research, are renowned for creating intricate and ever-changing spatial experiences within confined areas. Despite the existing literature and theories that attempt to explain these rich experiential qualities, many of these explanations need concrete empirical evidence due to the complex nature of gardens, where visual characteristics transform with the movement of people. This study employs a computational analysis method known as isovist to measure the evolving visual features of visitors along four representative pathways within a large-scale garden, the Humble Administrator’s Garden. By analyzing and comparing the changing visual attributes of these four routes, the aim is to validate the relationship between the garden’s pathway system and its spatial structure and assess the influence of pathway selection on the overall garden spatial experience.

Semantic segmentation of textured mosaics

This paper investigates deep learning (DL)-based semantic segmentation of textured mosaics. Existing popular datasets for mosaic texture segmentation, designed prior to the DL era, have several limitations: (1) training images are single-textured and thus differ from the multi-textured test images; (2) training and test textures are typically cut out from the same raw images, which may hinder model generalization; (3) each test image has its own limited set of training images, thus forcing an inefficient training of one model per test image from few data. We propose two texture segmentation datasets, based on the existing Outex and DTD datasets, that are suitable for training semantic segmentation networks and that address the above limitations: SemSegOutex focuses on materials acquired under controlled conditions, and SemSegDTD focuses on visual attributes of textures acquired in the wild. We also generate a synthetic version of SemSegOutex via texture synthesis that can be used in the same way as standard random data augmentation. Finally, we study the performance of the state-of-the-art DeepLabv3+ for textured mosaic segmentation, which is excellent for SemSegOutex and variable for SemSegDTD. Our datasets allow us to analyze results according to the type of material, visual attributes, various image acquisition artifacts, and natural versus synthetic aspects, yielding new insights into the possible usage of recent DL technologies for texture analysis.

Discriminative and Robust Attribute Alignment for Zero-Shot Learning

Zero-shot learning (ZSL) aims to learn models that can recognize images of semantically related unseen categories, through transferring attribute-based knowledge learned from training data of seen classes to unseen testing data. As visual attributes play a vital role in ZSL, recent embedding-based methods usually focus on learning a compatibility function between the visual representation and the class semantic attributes. While in this work, in addition to simply learning the region embedding of different semantic attributes to maintain the generalization capability of the learned model, we further consider to improve the discrimination power of the learned visual features themselves by contrastive embedding. It exploits both the class-wise and instance-wise supervision for GZSL, under the attribute guided weakly supervised representation learning framework. To further improve the robustness of the ZSL model, we also propose to train the model under the consistency regularization constraint, through taking full advantages of self-supervised signals of the image under various perturbed augmentation situations, which could make the model robust to some occluded or un-related attribute regions. Extensive experimental results demonstrate the effectiveness of the proposed ZSL method, achieving superior performances to state-of-the-art methods on three widely-used benchmark datasets, namely CUB, SUN, and AWA2. Our source code is released at https://github.com/KORIYN/CC-ZSL.

Impact of the physicochemical parameters on the sensory characterization of Brazilian coffee by the CATA method

This research aims to evaluate the sensory description of arabica and canephora coffee by consumers, identify how and canephora coffee by consumers, and identify how physicochemical variables affect sensory perception. Three Brazilian coffees (two arabica and one canephora) were determined sensorial attributes by the Check-All-That-Apply (CATA) method with 190 consumers in a home test, and physicochemical (pH and caffeine content) and chromatic parameters. The results showed that visual perception, brightness, and orange coloration are the main means used to distinguish arabica coffee from commercial coffee. In addition, the perception of sweeter aromas (toffee, vanilla, and caramel spread), chocolate with greater intensity, and the less intense bitterness taste were important typicality markers to differentiate arabica coffee (cv. Typica). The analysis of the correlations between caffeine and bitterness, sweetness, and acidity demonstrates the consumer's ability to perceive basic tastes in the different coffee samples. The chromatic parameters obtained significant correlations with visual, olfactory, and gustatory attributes. This study indicated the consumer's ability to establish typical traits for Arabica coffee depending on the place of cultivation. In addition, consumers perceive the influence of physicochemical variables on sensory attributes, highlighting the need for quality in an increasingly demanding and globalized market.

Scatter of mechanical properties of 2D biaxial and triaxial braided reinforced composites and their correlation with visual features

Tensile properties of 2D braid reinforced composites are analyzed with regards to process-induced fiber architecture variations. First, a parametric study concerning the effect of radial braiding parameters on the braid characteristics is performed and a quantitative quality index based on visual attributes is introduced. Second, test laminates are produced with vacuum resin infusion to conduct tensile tests with biaxial and triaxial braids of varying qualities. To correlate fiber architecture variations with mechanical properties, each braided layer is scanned and analyzed by means of gray scale based image processing. This allows for determining specimen-specific fiber orientations and correlating them with mechanical properties. The results show strong correlations of measured fiber architectures and mechanical properties. Thus, the study suggests that detailed information about the (local) fiber architecture should allow for significantly improving the prediction of 2D braided composite properties.

Enhancing Shoppers’ Experiences and Building Mall Loyalty: The Role of Octomodal Mental Imagery (OMI) and Management Dimension-Evidence from the Yangtze River Delta Region of China

This study investigated the impact of Octomodal Mental Imagery (OMI) and management dimensions on shoppers’ experiences and mall loyalty. The study was collected through “Questionnaire Star,” and 358 valid data points were obtained. The data were analyzed using SPSS27 and AMOS28. The results showed that sensory attributes such as visual, tactile, gustatory, and olfactory attributes positively influenced shoppers’ experiences in OMI, while the auditory attribute had no significant effect. Spatial, a structural attribute of OMI, positively influenced shoppers’ experiences, whereas autonomy and kinesthetics did not have a considerable effect. Tenant mix and entertainment positively impacted shoppers’ experiences in management, while accessibility had no significant effect. The study also found that shoppers’ experiences positively impacted mall loyalty, while hedonistic motivation had a more substantial effect than utilitarian motivation. This study is the first to examine the impact of OMI on shoppers’ experiences. It fills a gap in the literature on this relationship. It also examines the combined impact of management dimensions (accessibility, tenant mix, and entertainment) on the overall shopper’s experience, filling a gap in the Chinese shopping mall literature and extending the generalizability of the theory. The study further explores the relationship between shoppers’ experiences and mall loyalty and the moderating effect of incentive orientation. The results of this study have critical implications for mall managers. Strengthening the mental image and management dimensions of the shopping mall will enhance shoppers’ experiences and build loyalty, allowing brick-and-mortar malls to remain competitive and sustainable in today’s highly competitive and popular e-commerce environment.

Waterways transformation in the vulnerable port city of Alexandria

The IPCC AR6 report advises that climate change will exacerbate storm surges and coastal flooding in the eastern Mediterranean basin in the upcoming decades. The above is particularly acute in low-lying arid cities in developing nations like Alexandria. The communities' lack of awareness of these rising vulnerabilities subsequently impacts their willingness to tackle upcoming challenges. The latter strongly relies on the public perception and efficiency of the ongoing waterways and waterscape transformations, which remains largely unquantified. To address this deficiency, we evaluate these factors for the historic transformed 21-km canal in Alexandria's vulnerable arid port city. We perform a public survey based on the City Development Strategy for Sustainable Development criterion, assessing the inhabitants' perceptions of the degeneration and transformation of the city's most iconic and historic urban waterway. Our results suggest that the waterscape's multiple transformations negatively influenced the inhabitants' awareness of Alexandria's coastal vulnerabilities and led to public disconnection from the city's maritime heritage, visual attributes, and environment. This disconnection resulting from decades of degrading waterways and inefficient landscape transformations is mainly observed among youth and further compromises the city's resilience to upcoming coastal hazards and aggravates the ecological and coastal degradations. We suggest an adaptive conceptual plan of waterways regeneration to restore their roles in mitigating climatic challenges and improving public connectivity to the historic city's marine identity.

Learning Semantic Representation on Visual Attribute Graph for Person Re-identification and Beyond

Person re-identification (re-ID) aims to match pedestrian pairs captured from different cameras. Recently, various attribute-based models have been proposed to combine the pedestrian attribute as an auxiliary semantic information to learn a more discriminative pedestrian representation. However, these methods usually directly concatenate the visual branch and attribute branch embeddings as the final pedestrian representation, which ignores the semantic relation between the pedestrian revealed by attribute similarity. To capture and explore such semantic relation, we propose a unified pedestrian representation framework, called Visual Attribute Graph Embedding Network (VAGEN), to simultaneously learn attribute and visual representation. We unify the visual embedding and attribute similarity into a Visual Attribute Graph, where pedestrian is considered as a node and attribute similarity as an edge. Then, we learn graph node embedding to generate pedestrian representation through Graph Neural Network. Except for this unified representation for visual and attribute embeddings, VAGEN also conducts implicitly hard example mining for visual similar false-positive results, which has not been explored yet among existing attribute-based methods. We conduct extensive empirical studies on several person re-ID datasets to evaluate our proposed algorithm from different aspects. The results show that our proposed method outperforms state-of-the-art techniques with considerable margins.

PerceptSent - Exploring Subjectivity in a Novel Dataset for Visual Sentiment Analysis

Visual sentiment analysis is a challenging problem. Many datasets and approaches have been designed to foster breakthroughs in this trending research topic. However, most works scrutinize only subsymbolic models through visual attributes of the evaluated images, paying less attention to the subjectivity of viewers' perceptions as a basis for neuro-symbolic systems. Aiming to fill this gap, we present PerceptSent, a novel dataset for visual sentiment analysis that spans 5,000 images shared by users on social networks. Besides the sentiment opinion (positive, slightly positive, neutral, slightly negative, negative) expressed by every evaluator about each image analyzed, the dataset contains evaluator's metadata (age, gender, socioeconomic status, education, and psychological hints) as well as perceptions observed by the evaluator about the image — such as the presence of nature, violence, lack of maintenance, etc. Deep architectures and different problem formulations are explored using our dataset to combine visual and extra attributes (external knowledge) for automatic sentiment analysis. We show evidence that evaluator's perceptionss, when correctly employed, are crucial in visual sentiment analysis, improving the F-score performance from 61% to an impressive rate above 97%. Although, at this point, we do not have automatic approaches to capture these perceptions, our results open up new investigation avenues.

A personalized bikeability-based cycling route recommendation method with machine learning

Urban built environment quality along the travel route is positively correlated with active mobility. However, most previous personalized route recommendation research for cycling emphasizes non-visual attributes, ignoring the significance of visual attributes such as the built environment. Therefore, to take the built environment into consideration in the cycling route planning, we propose the personalized bikeability-based cycling route recommendation (PBCRR) method. The method consists of three steps: (1) deep neural networks are utilized to achieve statistics on built environment attributes and features of bikeability; (2) the weights road impedance calculation function is used to calculate route recommendation indicators according to minimum impedance sum of road segments; and (3) the recommended cycling route is found and shown to the user based on the calculated indicators. The results indicate that our PBCRR presents a high level of effectiveness and differences, and achieves better results than the current route recommendation algorithm in guiding cyclists towards routes with a better built environment.

SPRING: Situated Conversation Agent Pretrained with Multimodal Questions from Incremental Layout Graph

Existing multimodal conversation agents have shown impressive abilities to locate absolute positions or retrieve attributes in simple scenarios, but they fail to perform well when complex relative positions and information alignments are involved, which poses a bottleneck in response quality. In this paper, we propose a Situated Conversation Agent Pretrained with Multimodal Questions from Incremental Layout Graph (SPRING) with abilities of reasoning multi-hops spatial relations and connecting them with visual attributes in crowded situated scenarios. Specifically, we design two types of Multimodal Question Answering (MQA) tasks to pretrain the agent. All QA pairs utilized during pretraining are generated from novel Increment Layout Graphs (ILG). QA pair difficulty labels automatically annotated by ILG are used to promote MQA-based Curriculum Learning. Experimental results verify the SPRING's effectiveness, showing that it significantly outperforms state-of-the-art approaches on both SIMMC 1.0 and SIMMC 2.0 datasets. We release our code and data at https://github.com/LYX0501/SPRING.

How should B2B firms create image content for high social media engagement? A multimodal analysis

PurposeThis study evaluates the effect of the media image content of business to business (B2B) organizations to accelerate social media engagement. It highlights the importance of strategically designing image content for business marketing strategies.Design/methodology/approachThis study designed a computation extensive research model based upon the stimulus-organism-response (SOR) theory using 39,139 Facebook posts of 125 organizations selected from Fortune 500 firms. Attributes from images and text were estimated using deep learning models. Subsequently, inferential analysis was established with ordinary least squares regression. Further machine learning algorithms, like support vector regression, k-nearest neighbour, decision tree and random forest, are used to analyze the significance and robustness of the proposed model for predicting engagement metrics.FindingsThe results indicate that the social media (SM) image content of B2B firms significantly impacts their social media engagement. The visual and linguistic attributes are extracted from the image using deep learning. The distinctive effect of each feature on social media engagement (SME) is empirically verified in this study.Originality/valueThis research presents practical insights formulated by embedding marketing, advertising, image processing and statistical knowledge of SM analytics. The findings of this study provide evidence for the stimulating effect of image content concerning SME. Based on the theoretical implications of this study, marketing and media content practitioners can enhance the efficacy of SM posts in engaging users.

EXAMINING THE BOUNDARY BETWEEN NEAR AND FAR DESIGN STIMULI

AbstractExternal sources of inspiration can promote the discovery of new ideas as designers ideate on a design task. Data-driven techniques can increasingly enable the retrieval of inspirational stimuli based on nontext-based representations, beyond semantic features of stimuli. However, there is a lack of fundamental understanding regarding how humans evaluate similarity between non-semantic design stimuli (e.g., visual). Toward this aim, this work examines human-evaluated and computationally derived representations of visual and functional similarities of 3D-model parts. A study was conducted where participants (n=36) assessed triplet ratings of parts and categorized these parts into groups. Similarity is defined by distances within embedding spaces constructed using triplet ratings and deep-learning methods, representing human and computational representations. Distances between stimuli that are grouped together (or not) are determined to understand how various methods and criteria used to define non-text-based similarity align with perceptions of 'near' and 'far'. Distinct boundaries in computed distances separating stimuli that are 'too far' were observed, which include farther stimuli when modeling visual vs. functional attributes.

A novel content-based image retrieval system with feature descriptor integration and accuracy noise reduction

Efficient algorithms, intelligent approaches, and intensive use of computers are crucial to extracting reliable information from large-scale data sets. It is the case, for example, in retrieving information from multimedia data. As different systems integrate individual electronic devices such as smartphones and cameras into storage, sharing, and social media platforms, the amount of multimedia data has increased dramatically in recent years. Therefore, content-based image retrieval (CBIR) is quite challenging. In this paper, we introduce a new image descriptors integration to represent the visual attributes of images. Furthermore, we noticed a pattern between different CBIR systems in which the first images retrieved are more likely to be assertive than those in the middle and final positions. Then, we investigated the interactions between the first images retrieved from CBIR systems to present a novel method for reducing accuracy noise. The performance of the proposed method is evaluated in different data sets (Corel-1K, Corel-5K, Corel-10K, GHIM-10K) and compared to related works. The experimental results demonstrate that the proposal consistently retrieves images with similar content. The code prepared by the authors is publicly available.