Caption Text Research Articles

Corrigendum.

American Journal of BotanyVolume 107, Issue 6 p. 950-950 CorrigendumFree Access Corrigendum This article corrects the following: A consensus phylogenomic approach highlights paleopolyploid and rapid radiation in the history of Ericales Drew A. Larson, Joseph F. Walker, Oscar M. Vargas, Stephen A. Smith, Volume 107Issue 5American Journal of Botany pages: 773-789 First Published online: April 29, 2020 First published: 26 June 2020 https://doi.org/10.1002/ajb2.1503AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinked InRedditWechat Due to a graphical error, Figure 6 contains a phylogeny that incorrectly depicts one phylogenetic relationship in the consensus topology. The analyses and descriptions of the consensus topology in the caption and main text are correct, and the error does not affect the main message of the figure. The figure is corrected here. –The Authors. Reference Larson, D. A., J. F. Walker, O. M. Vargas, and S. A. Smith. 2020. A consensus phylogenomic approach highlights paleopolyploid and rapid radiation in the history of Ericales. American Journal of Botany 107(5): 773– 789Wiley Online LibraryPubMedWeb of Science®Google Scholar Volume107, Issue6June 2020Pages 950-950 ReferencesRelatedInformation

Detecting Gurmukhi and Hindi text object present in images

This paper resolves the trouble detecting Gurmukhi and Hindi text objects present in images. Therefore, we propose a new region-based bottom-up method to extract text from images, where we first identify elementary substructures using connected components and edges and then merge them successively. As each character counter has high contrast as compared to its neighbours, all character pixels and several non-character pixels which exhibit excessive neighbourhood. Although a lot of noise is removed by this step a second pass filter is applied based on stroke width. The stroke width of Hindi and Gurmukhi characters are approximately uniform. The candidates with similar properties are aggregated into chains, according to the observation that the figures of a book. For performance evaluation, we created our own dataset containing 60 images of varied size, resolution and type. The performance is evaluated on this dataset containing caption, document, and scene text images using different performance metrics.

Detecting Gurmukhi and Hindi text object present in images

This paper resolves the trouble detecting Gurmukhi and Hindi text objects present in images. Therefore, we propose a new region-based bottom-up method to extract text from images, where we first identify elementary substructures using connected components and edges and then merge them successively. As each character counter has high contrast as compared to its neighbours, all character pixels and several non-character pixels which exhibit excessive neighbourhood. Although a lot of noise is removed by this step a second pass filter is applied based on stroke width. The stroke width of Hindi and Gurmukhi characters are approximately uniform. The candidates with similar properties are aggregated into chains, according to the observation that the figures of a book. For performance evaluation, we created our own dataset containing 60 images of varied size, resolution and type. The performance is evaluated on this dataset containing caption, document, and scene text images using different performance metrics.

Использование синтетических данных для тонкой настройки моделей сегментации документов

In this paper, we propose an approach to the document images segmentation in a case of limited set of real data for training. The main idea of our approach is to use artificially created data for training and post-processing. The domain of the paper is PDF documents, such as scanned contracts, commercial proposals and technical specifications without a text layer is considered as data. As part of the task of automatic document analysis, we solve the problem of segmentation of DLA documents (Document Layout Analysis). In the paper we train the known high-level FasterRCNN \cite{ren2015faster} model to segment text blocks, tables, stamps and captions on images of the domain. The aim of the paper is to generate synthetic data similar to real data of the domain. It is necessary because the model needs a large dataset for training and the high labor intensity of their preparation. In the paper, we describe the post-processing stage to eliminate artifacts that are obtained as a result of the segmentation. We tested and compared the quality of a model trained on different datasets (with / without synthetic data, small / large set of real data, with / without post-processing stage). As a result, we show that the generation of synthetic data and the use of post-processing increase the quality of the model with a small real training data.

The Presentation of Jerome’s First Letter to Paulinus of Nola in the Codex Amiatinus Pentateuch Diagram

A diagram in the first quire of the Codex Amiatinus features five textual captions arranged in cruciform formation, one for each book of the Pentateuch. These are taken from Jerome’s first letter to Paulinus of Nola (Epistle 53) which was written in 394 ad. This article examines the diagram’s colours, geometric structure, manuscript location and script. It suggests that the Pentateuch diagram should be regarded as a highly original piece of visual exegesis designed to celebrate the contribution made by Jerome to the transmission of the Latin Bible and point the viewer towards typological interpretations of Old Testament figures and events.

Clustering User Characteristics Based on the influence of Hashtags on the Instagram Platform

Instagram is a social media that has the potential to be used to increase awareness of a product. Approximately 70% of users spend their time searching for a product on Instagram. Many people promote their products with a lack of attention to the target. So that not infrequently the information distributed is inaccurate information and not following user characteristics. This study aims to cluster the characteristics of Instagram users based on hashtag compatibility. The method used in this study is the K-Means Clustering method. Based on the results of the experiment, this research succeeded in clustering Instagram users based on the hashtag match on the text caption. Besides, TF-IDF can be used as a feature suitable for the K-Means Klastering method. The results of the hashtag "#kopi" analysis resulted in hashtag suggestions that can be used for the promotion of a product related to coffee, including the hashtag #coffeeshop and #coffee with total usage of 14968 captions.

An Efficient Technique for Scene Text Extraction from Videos

Text Detection from natural scene images and videos is imperative for applications in real world domain analysis. However the text detection process isperplexingbecause of exigent scenarios that the text exhibit. The information present in the video is either perceptual or it is either in semantic form. Amongst the different content that exists in the video, the text information is a major important content that describes more about the nature of the video. The text present in the video can be categorized into Caption text and Scene text. The caption text is the artificial text that is easy to detect while scene text are natural text which is difficult to identify. In this paper text extraction in natural images by edge based method is implemented. The algorithms are estimated with a set of images of natural scenes that differ alongside the scope of font size, illumination, scale and text direction. Precision, accuracy and recall rates are determined to evaluate the performance. The proposed system worked for all difficult scenarios of varied text and gave better results than the existing methods.

Computer Interface as Film: Post-Media Aesthetics of Desktop Documentary

This paper explores a recently emerged audiovisual form called desktop documentary, an interdisciplinary computer-based variant of the essay film. As a post-media practice, no longer exclusively dependent on the film medium, desktop filmmaking represents a hybrid audiovisual genre entirely conducted in the digital environment by using and exploiting preexisting materials in new contexts while using the advantages of the Internet, widely used software and digital tools. Desktop documentary filmmaking corresponds to the widespread artistic practice of postproduction – a concept introduced by Nicolas Bourriaud signifying a new state of affairs when all texts of culture are already available (mostly as digital objects) and the artist intervenes on existing materials rather than produces artworks ex nihilo. Belonging to the tradition of essay film – a cinematic documentary and experimental mode in which moving images and off-screen verbal voice or textual captions establish complex relations – desktop video essays introduce new post-media aesthetics. Similar to the idea of using everyday materials in the artistic context, initially proposed with Duchamp’s ready-mades, which unprecedentedly effaced every notion of the style from their avant-garde aesthetics, desktop documentaries often minimize and abolish cinematic stylistic qualities. One of the most significant aspects of desktop documentaries is that the act of film viewing does not differ from common computer user experience: having replaced traditional film screen with the computer interface, the interactive process of computational multitasking and navigation, performed on various digital data and files, becomes the very content of the film. After the historical overview of the phenomenon and general introduction into the post-media theory, selected works of representative desktop documentarists such as Kevin B. Lee and Louis Henderson are analyzed in their deconstructive approach to traditional and digital filmmaking – subversive both formally and politically. Article received: May 28, 2019; Article accepted: July 6, 2019; Published online: October 15, 2019; Original scholarly paper

Traditional ethnocultural communities in Soviet cinema: «Hunting and reindeer breeding in the Komi region»

Introduction. Visually-anthropological experiments in Russia have more than a century of history, reflected, in particular, in the corpora of Soviet ethnographic films in the 1920s – 1930s. This article, based on textual and visual materials, introduces information about the ethno-film «Hunting and Reindeer Breeding in the Komi Region» (1927) as a multi-layered visual-anthropological document. The paper considers the specific features of film history on the evolution among Komi hunters and reindeer herders during the period of cultural and economic transformations in the USSR in the second half of the 1920s, created in the mentioned film, in the context of the Soviet film production. Materials and Methods. Due to the specifics of silent cinema, the final film «Hunting and Reindeer Breeding in the Komi Region» is a kind of cinema-text consisting of approximately the same number of film frames and text captions interspersed in the narration. That is why the method of analyzing the film as a visual-text work was its research transcript – presentation in the form of a film text. Results and Discussion. Based on the archival data and published works of contemporaries, the authors traced the evolution of the film project in connection with parallel processes in the state national-cultural policy and in ethnographic science. Conclusion. Considering the film in the socio-historical context, the authors conclude the phenomenon of ethnographic film as an effective form of research knowledge, which allows fixing the cultures of what is being filmed and shooting, and broadcast in time not only actual events, but also their so important for anthropological study emotional context; as well as the potential of cinema as an informative historical source.

Recall What You See Continually Using GridLSTM in Image Captioning

The goal of image captioning is to automatically describe an image with a sentence, and the task has attracted research attention from both the computer vision and natural-language processing research communities. The existing encoder–decoder model and its variants, which are the most popular models for image captioning, use the image features in three ways: first, they inject the encoded image features into the decoder only once at the initial step, which does not enable the rich image content to be explored sufficiently while gradually generating a text caption; second, they concatenate the encoded image features with text as extra inputs at every step, which introduces unnecessary noise; and, third, they using an attention mechanism, which increases the computational complexity due to the introduction of extra neural nets to identify the attention regions. Different from the existing methods, in this paper, we propose a novel network, Recall Network, for generating captions that are consistent with the images. The recall network selectively involves the visual features by using a GridLSTM and, thus, is able to recall image contents while generating each word. By importing the visual information as the latent memory along the depth dimension LSTM, the decoder is able to admit the visual features dynamically through the inherent LSTM structure without adding any extra neural nets or parameters. The Recall Network efficiently prevents the decoder from deviating from the original image content. To verify the efficiency of our model, we conducted exhaustive experiments on full and dense image captioning. The experimental results clearly demonstrate that our recall network outperforms the conventional encoder–decoder model by a large margin and that it performs comparably to the state-of-the-art methods.

Learning semantic sentence representations from visually grounded language without lexical knowledge

AbstractCurrent approaches to learning semantic representations of sentences often use prior word-level knowledge. The current study aims to leverage visual information in order to capture sentence level semantics without the need for word embeddings. We use a multimodal sentence encoder trained on a corpus of images with matching text captions to produce visually grounded sentence embeddings. Deep Neural Networks are trained to map the two modalities to a common embedding space such that for an image the corresponding caption can be retrieved and vice versa. We show that our model achieves results comparable to the current state of the art on two popular image-caption retrieval benchmark datasets: Microsoft Common Objects in Context (MSCOCO) and Flickr8k. We evaluate the semantic content of the resulting sentence embeddings using the data from the Semantic Textual Similarity (STS) benchmark task and show that the multimodal embeddings correlate well with human semantic similarity judgements. The system achieves state-of-the-art results on several of these benchmarks, which shows that a system trained solely on multimodal data, without assuming any word representations, is able to capture sentence level semantics. Importantly, this result shows that we do not need prior knowledge of lexical level semantics in order to model sentence level semantics. These findings demonstrate the importance of visual information in semantics.

It\u2019s not raining men: a mixed-methods study investigating methods of improving male recruitment to health behaviour research

BackgroundAlthough gender is an important determinant of health behaviour with males less likely to perform health-protective behaviours, samples in health behaviour research are heavily biased towards females. This study investigated the use of online social network, Facebook, to reach and recruit inactive males to a team-based, social, and gamified physical activity randomised controlled trial.MethodsMethodological techniques included a narrative literature review, survey of inactive males (n = 34) who rated advertisement images and text captions on scales of 1–10, and trial Facebook-delivered recruitment campaigns. Advertisement effectiveness was measured by cost-per-click to the study website, number of expressions of interest, and study enrolments from males.ResultsSurvey results showed that vibrant images of men exercising accompanied by concise captions (< 35 words) were most effective. An advertising campaign incorporating these components achieved a cost-per-click of $0.60, with 80% of n = 50 expressions of interest being from men, a marked improvement from baseline campaigns in which only 11% of expressions of interest were from men. Despite this, men who were recruited through the targeted campaign failed to enrol into the study, primarily due to reluctance to invite friends to join their team. An alternative strategy of encouraging females to invite men boosted male participation from 18% of the sample at baseline to 29% in the targeted recruitment phase.ConclusionsEvidence-based approaches can improve Facebook recruitment outcomes, however, there are complex barriers hindering male recruitment to health behaviour studies that may necessitate multi-faceted strategies including involvement of family and friends.

A Real-time Character-level Subtitle Detector based on Faster R-CNN

Faced with a deluge of video data, automatic subtitle text detection and recognition using computer vision technology can be extremely useful. This paper proposes a real-time character-level subtitle detector, which detects subtitle text with high accuracy and efficiency in a single forward pass. It also provides an efficient method to build dataset aimed at video subtitle text detection problem. This paper’s method can achieve both 99.5% precision and recall for text detection, and the whole detection and recognition process can achieve 45 fps.

Readers’ and Photojournalists’ Perceptions of Print Media Road Carnage Images in The Herald

ABSTRACTThis study investigates possible semiotic and linguistic effects of road carnage visual images in print media, as perceived and conceptualised, respectively, by readers and photojournalists of The Herald newspaper in Zimbabwe in the context of the government’s public campaign on the prevention of the high incidence of fatal road accidents. It proposes that images have transcended the monotony of texts and could now tell the same story, sometimes, much better. The study, therefore, sought to: establish the onset of visual news as a social semiotic meaning-making process in the print media; establish ethical and cultural considerations in reading print media visual images; examine the use of framing in creating news through visual images; and, analyse the combined impact of visuals and text captions in news reading. Using visual image content analysis (n=43); a questionnaire for regular readers of the newspaper (n=100) and photojournalists’ interviews (n=5), the study concludes that visual media images, specifically road carnage images from The Herald, are a form of “language” that is fast becoming a new mode of effective communication in emotion-inducing news. Similar to Austin’s theory on speech acts, the reading of road carnage visuals should be treated as a visual act.

Protagonist and deuteragonist-based video indexing and retrieval system for movie and video song sequences

Protagonist and deuteragonist are two main characters that plays leading role in Indian Hindi movie (IHM). Currently such information is attached using textual caption which is highly unreliable. The research presented in this paper is to automatically index and retrieve the content based on protagonist and deuteragonist from the large IHM and video song sequences (VSS). Video song sequence indexes are extracted using audio-based approach from IHM in module 1. These indexes are used as an input in module 2. Faces are identified from every VSS. Colour histogram and spatiogram descriptor are extracted from faces. Similarity between two faces is computed using Bhattacharyya coefficient. Similarity-based clustering technique is performed to obtain clusters of faces. Recognition of protagonist and deuteragonist is done using surf feature points. The experimental results are carried out using Indian Hindi movies of different genres.

Этнокультурные сообщества Севера в этнокино «У берегов Чукотского моря» Григория Смирнитского: история съемок

Purpose. History of visual anthropology in Russia is full of big names, heroic expeditions and classic films. Based on research and pursuing artistic goals, production of ethnographic films was also part of a grand Soviet experiment on creating a unique national identity. Soviet authorities used popular “cultural films” as a means of bringing together the peoples of the new Soviet Union on the screen. This article deals with exploring the film-image of the culture of the Chukchi on the example of a classic documentary film shot by Grigory Smirnitsky, a pioneer of the Soviet visual anthropology. The film was a result of the Chukotka expedition conducted by the film group of the Soyuzkino factory under the leadership of A. Litvinov, one of pioneers and icons of the national ethnographic cinema. The film “Coast of the Chukchi Sea”, the main result of the film expedition which has survived up to our days, was created by an A. Litvinov’s assistant, a young director and scriptwriter G. Smirnitsky. Due to the specifics of silent movies, this film is a kind of a cinematic text as it consists of approximately the same number of frames and text captions alternating in the narrative. In this regard, an effective method for analyzing this film applied in this article is its decoding, a «translation» into a text format. Result. The resultant film-text allows us to identify and analyze, on the one hand, the features of the screen image of the Chukchi culture of the early twentieth century. On the other hand, we analyzed the basis of the specific creative method of the director-researcher. G. Smirnitsky’s creativity is a perfect example of how a film potential can be used as a form of research cognition. The method discovered by the director in the expedition helps to combine research and documentary elements into a film so that the film could convey not only some information about the events shown, but also their figurative and emotional context. Conclusion. Based on the analysis of visual, textual and archival materials, we come to the conclusion that this documentary film phenomenon is a valuable historical source. Studying Soviet ethnographic cinema, including the film “Coast of the Chukchi Sea”, enriches the source base of modern science with audiovisual ethnographic materials and equips scientists with proven scientific and creative methods that could be used in modern research practice.

Learning deep spatiotemporal features for video captioning

In this paper, we propose a novel automatic video captioning system which translates videos to sentences, utilizing a deep neural network that is composed of three building parts of convolutional and recurrent structure. That is, the first subnetwork operates as feature extractor of single frames. The second subnetwork is a three-stream network, capable of capturing spatial semantic information in the first stream, temporal semantic information in the second stream, and global video concept information in the third stream. The third subnetwork generates relevant textual captions using as input the spatiotemporal features of the second subnetwork. The experimental validation indicates the effectiveness of the proposed model, achieving superior performance over competitive methods.

Extracting brand information from social networks: Integrating image, text, and social tagging data

Images are an essential feature of many social networking services, such as Facebook, Instagram, and Twitter. Through brand-related images, consumers communicate about brands with each other and link the brand with rich contextual and consumption experiences. However, previous articles in marketing research have concentrated on deriving brand information from textual user-generated content and have largely not considered brand-related images. The analysis of brand-related images yields at least two challenges. First, the content displayed in images is heterogeneous, and second, images rarely show what users think and feel in or about the situations displayed. To meet these challenges, this article presents a two-step approach that involves collecting, labeling, clustering, aggregating, mapping, and analyzing brand-related user-generated content. The collected data are brand-related images, caption texts, and social tags posted on Instagram. Clustering images labeled via Google Cloud Vision API enabled to identify heterogeneous contents (e.g. products) and contexts (e.g. situations) that consumers create content about. Aggregating and mapping the textual information for the resulting image clusters in the form of associative networks empowers marketers to derive meaningful insights by inferring what consumers think and feel about their brand regarding different contents and contexts.

Instagram abroad: performance, consumption and colonial narrative in tourism

ABSTRACTAs a form of travel writing and a highly favoured marketing tool, Instagram provides a blueprint of the ideologies underpinning contemporary tourism. This article argues that consistent visual motifs on Instagram echo a colonial iconography that sees tourist destinations as available for possession and consumption, effacing local place and identity. The reproduction of three motifs – the tropical exotic, the promontory gaze and fantasised assimilation – mediatises ideations that, rather than depicting these destinations as contemporaneous spaces in which a tourist is a guest, depict them as ‘other’ realms for the tourist’s taking. Local residents, when pictured, are configured as genericised icons of exoticism that serve to imbue the tourist’s experience with authenticity. These visual tropes, paired with textual captions and hashtags, present tourists as the rightful occupants and users of local spaces in a way that echoes the colonial seizure of foreign lands, an action that is imaginatively performed as tourists enact these three motifs in Instagram posts. Taken together, the visual regime witnessed and performed on Instagram contributes to the imagined and real perpetuation of unequal power relations in global tourism, which continue to privilege wealthy tourists over local residents.

Old jokes, new media – Online sexism and constructions of gender in Internet memes

The Internet is a space where the harassment of women and marginalised groups online has attracted the attention of both academic and popular press. Feminist research has found that instances of online sexism and harassment are often reframed as “acceptable” by constructing them as a form of humour. Following this earlier research, this present paper explores a uniquely technologically-bound type of humour by adopting a feminist, social-constructionist approach to examine the content of popular Internet memes. Using thematic analysis on a sample of 240 image macro Internet memes (those featuring an image with a text caption overlaid), we identified two broad, overarching themes – Technological Privilege and Others. Within the analysis presented here, complex and troubling constructions of gendered identity in online humour are explored, illustrating the potential for the othering and exclusion of women through humour in technological spaces. We argue that this new iteration of heteronormative, hegemonic masculinity in online sexism, couched in “irony” and “joking”, serves to police, regulate and create rightful occupants and owners of such spaces.