Visual Captioning Research Articles

ZeroNLG: Aligning and Autoencoding Domains for Zero-Shot Multimodal and Multilingual Natural Language Generation.

Natural Language Generation (NLG) accepts input data in the form of images, videos, or text and generates corresponding natural language text as output. Existing NLG methods mainly adopt a supervised approach and rely heavily on coupled data-to-text pairs. However, for many targeted scenarios and for non-English languages, sufficient quantities of labeled data are often not available. As a result, it is necessary to collect and label data-text pairs for training, which is both costly and time-consuming. To relax the dependency on labeled data of downstream tasks, we propose an intuitive and effective zero-shot learning framework, ZeroNLG, which can deal with multiple NLG tasks, including image-to-text (image captioning), video-to-text (video captioning), and text-to-text (neural machine translation), across English, Chinese, German, and French within a unified framework. ZeroNLG does not require any labeled downstream pairs for training. During training, ZeroNLG (i) projects different domains (across modalities and languages) to corresponding coordinates in a shared common latent space; (ii) bridges different domains by aligning their corresponding coordinates in this space; and (iii) builds an unsupervised multilingual auto-encoder to learn to generate text by reconstructing the input text given its coordinate in shared latent space. Consequently, during inference, based on the data-to-text pipeline, ZeroNLG can generate target sentences across different languages given the coordinate of input data in the common space. Within this unified framework, given visual (imaging or video) data as input, ZeroNLG can perform zero-shot visual captioning; given textual sentences as input, ZeroNLG can perform zero-shot machine translation. We present the results of extensive experiments on twelve NLG tasks, showing that, without using any labeled downstream pairs for training, ZeroNLG generates high-quality and "believable" outputs and significantly outperforms existing zero-shot methods.

Visual Chain-of-Thought Prompting for Knowledge-Based Visual Reasoning

Knowledge-based visual reasoning remains a daunting task since it not only requires machines to interpret the concepts and relationships from visual scenes but also associate them with external world knowledge to conduct a chain of reasoning on open-world questions. Previous works, however, treat visual perception and language-based reasoning as two independent modules, failing to attend to both modules throughout all stages of reasoning. To this end, we propose Visual Chain-of-thought Prompting (VCTP) for knowledge-based reasoning, which involves the interaction between visual content and natural language in an iterative step-by-step reasoning manner. VCTP contains three stages, see, think, and confirm. The see stage scans the image and grounds the visual concept candidates with a visual perception model. The think stage adopts a pre-trained large language model (LLM) to attend to key visual concepts from natural language questions adaptively. It then transforms key visual context into text context for prompting with a visual captioning model, and adopts the LLM to generate the answer. The confirm stage further uses the LLM to generate the supporting rationale to the answer, which is then passed through a cross-modality classifier to verify that it’s consistent with the visual context. We iterate through the think-confirm stages to ensure the verified rationale is consistent with the answer. We conduct experiments on a range of knowledge-based visual reasoning datasets. We found our VCTP enjoys several benefits, 1). it achieves better performance than the previous few-shot learning baselines; 2). it enjoys the total transparency and trustworthiness of the whole reasoning process by providing rationales for each reasoning step; 3). it is computation-efficient compared with other fine-tuning baselines. Our code is available at https://github.com/UMass-Foundation-Model/VisualCoT.git

A comprehensive survey on deep-learning-based visual captioning

A comprehensive survey on deep-learning-based visual captioning

A Survey of Vision and Language Related Multi-Modal Task

With the significant breakthrough in the research of single-modal related deep learning tasks, more and more works begin to focus on multi-modal tasks. Multi-modal tasks usually involve more than one different modalities, and a modality represents a type of behavior or state. Common multi-modal information includes vision, hearing, language, touch, and smell. Vision and language are two of the most common modalities in human daily life, and many typical multi-modal tasks focus on these two modalities, such as visual captioning and visual grounding. In this paper, we conduct in-depth research on typical tasks of vision and language from the perspectives of generation, analysis, and reasoning. First, the analysis and summary with the typical tasks and some pretty classical methods are introduced, which will be generalized from the aspects of different algorithmic concerns, and be further discussed frequently used datasets and metrics. Then, some other variant tasks and cutting-edge tasks are briefly summarized to build a more comprehensive vision and language related multi-modal tasks framework. Finally, we further discuss the development of pre-training related research and make an outlook for future research. We hope this survey can help relevant researchers to understand the latest progress, existing problems, and exploration directions of vision and language multi-modal related tasks, and provide guidance for future research.

Affective word embedding in affective explanation generation for fine art paintings

Fine art paintings take an important place in human history and are one of the fundamental components of human culture. Even though deep learning in fine art paintings attracts increasing attention from researchers, few works focus on understanding the interplay between the visual content, its triggered affect, and language aiming to explain that affect. Most visual captioning models can only deal with tasks of generating captions describing objective affairs but lack the capabilities of generating affective explanations. In this paper, we introduce the use of the VAD (Valence, Arousal, and Dominance) dictionary in our model and propose a gated concatenation mechanism to construct word affective embedding. Corporating with the use of the affective loss function, our model outperforms the state-of-the-art in automatic evaluation metrics and subjective evaluations.

Sentiment Analysis of Image with Text Caption using Deep Learning Techniques.

People are actively expressing their views and opinions via the use of visual pictures and text captions on social media platforms, rather than just publishing them in plain text as a consequence of technical improvements in this field. With the advent of visual media such as images, videos, and GIFs, research on the subject of sentiment analysis has expanded to encompass the study of social interaction and opinion prediction via the use of visuals. Researchers have focused their efforts on understanding social interaction and opinion prediction via the use of images, such as photographs, films, and animated GIFs (graphics interchange formats). The results of various individual studies have resulted in important advancements being achieved in the disciplines of text sentiment analysis and image sentiment analysis. It is recommended that future studies investigate the combination of picture sentiment analysis and text captions in more depth, and further research is necessary for this field. An intermodal analysis technique known as deep learning-based intermodal (DLBI) analysis is discussed in this suggested study, which may be used to show the link between words and pictures in a variety of scenarios. It is feasible to gather opinion information in numerical vector form by using the VGG network. Afterward, the information is transformed into a mapping procedure. It is necessary to predict future views based on the information vectors that have been obtained thus far, and this is accomplished through the use of active deep learning. A series of simulation tests are being conducted to put the proposed mode of operation to the test. When we look at the findings of this research, it is possible to infer that the model outperforms and delivers a better solution with more accuracy and precision, as well as reduced latency and an error rate, when compared to the alternative model (the choice).

Semantic context driven language descriptions of videos using deep neural network

The massive addition of data to the internet in text, images, and videos made computer vision-based tasks challenging in the big data domain. Recent exploration of video data and progress in visual information captioning has been an arduous task in computer vision. Visual captioning is attributable to integrating visual information with natural language descriptions. This paper proposes an encoder-decoder framework with a 2D-Convolutional Neural Network (CNN) model and layered Long Short Term Memory (LSTM) as the encoder and an LSTM model integrated with an attention mechanism working as the decoder with a hybrid loss function. Visual feature vectors extracted from the video frames using a 2D-CNN model capture spatial features. Specifically, the visual feature vectors are fed into the layered LSTM to capture the temporal information. The attention mechanism enables the decoder to perceive and focus on relevant objects and correlate the visual context and language content for producing semantically correct captions. The visual features and GloVe word embeddings are input into the decoder to generate natural semantic descriptions for the videos. The performance of the proposed framework is evaluated on the video captioning benchmark dataset Microsoft Video Description (MSVD) using various well-known evaluation metrics. The experimental findings indicate that the suggested framework outperforms state-of-the-art techniques. Compared to the state-of-the-art research methods, the proposed model significantly increased all measures, B@1, B@2, B@3, B@4, METEOR, and CIDEr, with the score of 78.4, 64.8, 54.2, and 43.7, 32.3, and 70.7, respectively. The progression in all scores indicates a more excellent grasp of the context of the inputs, which results in more accurate caption prediction.

Imagine, Reason and Write: Visual Storytelling with Graph Knowledge and Relational Reasoning

Visual storytelling is a task of creating a short story based on photo streams. Different from visual captions, stories contain not only factual descriptions, but also imaginary concepts that do not appear in the images. In this paper, we propose a novel imagine-reason-write generation framework (IRW) for visual storytelling, inspired by the logic of humans when they write the story. First, an imagine module is leveraged to learn the imaginative storyline explicitly, improving the coherence and reasonability of the generated story. Second, we employ a reason module to fully exploit the external knowledge (commonsense knowledge base) and task-specific knowledge (scene graph and event graph) with relational reasoning method based on the storyline. In this way, we can effectively capture the most informative commonsense and visual relationships among objects in images, which enhances the diversity and informativeness of the generated story. Finally, we integrate the imaginary concepts and relational knowledge to generate human-like story based on the original semantics of images. Extensive experiments on a benchmark dataset (i.e., VIST) demonstrate that the proposed IRW framework significantly outperforms the state-of-the-art methods across multiple evaluation metrics.

Consensus Graph Representation Learning for Better Grounded Image Captioning

The contemporary visual captioning models frequently hallucinate objects that are not actually in a scene, due to the visual misclassification or over-reliance on priors that resulting in the semantic inconsistency between the visual information and the target lexical words. The most common way is to encourage the captioning model to dynamically link generated object words or phrases to appropriate regions of the image, i.e., the grounded image captioning (GIC). However, GIC utilizes an auxiliary task (grounding objects) that has not solved the key issue of object hallucination, i.e., the semantic inconsistency. In this paper, we take a novel perspective on the issue above: exploiting the semantic coherency between the visual and language modalities. Specifically, we propose the Consensus Rraph Representation Learning framework (CGRL) for GIC that incorporates a consensus representation into the grounded captioning pipeline. The consensus is learned by aligning the visual graph (e.g., scene graph) to the language graph that consider both the nodes and edges in a graph. With the aligned consensus, the captioning model can capture both the correct linguistic characteristics and visual relevance, and then grounding appropriate image regions further. We validate the effectiveness of our model, with a significant decline in object hallucination (-9% CHAIRi) on the Flickr30k Entities dataset. Besides, our CGRL also evaluated by several automatic metrics and human evaluation, the results indicate that the proposed approach can simultaneously improve the performance of image captioning (+2.9 Cider) and grounding (+2.3 F1LOC}).

Hierarchical Memory Decoder for Visual Narrating

Visual narrating focuses on generating semantic descriptions to summarize visual content of images or videos, e.g., visual captioning and visual storytelling. The challenge mainly lies in how to design a decoder to generate accurate descriptions matching visual content. Recent advances often employ a recurrent neural network (RNN), e.g., Long-Short Term Memory (LSTM), as the decoder. However, RNN is prone to diluting long-term information, which weakens its performance of capturing long-term dependencies. Recent work has demonstrated memory network (MemNet) owns the advantage of storing long-term information. However, as the decoder, it has not been well exploited for visual narrating. The reason partially comes from the difficulty of multi-modal sequential decoding with MemNet. In this article, we devise a novel memory decoder for visual narrating. Concretely, to obtain a better multi-modal representation, we first design a new multi-modal fusion method to fully merge visual and lexical information. Then, based on the fusion result, during decoding, we construct a MemNet-based decoder consisting of multiple memory layers. Particularly, in each layer, we employ a memory set to store previous decoding information and utilize an attention mechanism to adaptively select the information related to the current output. Meanwhile, we also employ a memory set to store the decoding output of each memory layer at the current time step and still utilize an attention mechanism to select the related information. Thus, this decoder alleviates dilution of long-term information. Meanwhile, the hierarchical architecture leverages the latent information of each layer, which is helpful for generating accurate descriptions. Experimental results on two tasks of visual narrating, i.e., video captioning and visual storytelling, show that our decoder could obtain superior results and outperform the performance of conventional RNN-based decoder.

Image captions: global-local and joint signals attention model (GL-JSAM)

For automated visual captioning, existing neural encoder-decoder methods commonly use a simple sequence-to-sequence or an attention-based mechanism. The attention-based models pay attention to specific visual areas or objects; using a single heat map that indicates which portion of the image is most important rather than treating the objects (within the image) equally. These models are usually a mixture of Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN) architectures. CNN’s generally extract global visual signals that only provide global information of main objects, attributes, and their relationship, but fail to provide local (regional) information within objects, such as lines, corners, curve and edges. On one hand, missing some of the information and details of local visual signals may lead to misprediction, misidentification of objects or completely missing the main object(s). On the other hand, additional visual signals information produces meaningless and irrelevant description, which may be coming from objects in foreground or background. To address these concerns, we created a new joint signals attention image captioning model for global and local signals that is adaptive by nature. Primarily, proposed model extracts global visual signals at image-level and local visual signals at object-level. The joint signal attention model (JSAM) plays a dual role in visual signal extraction and non-visual signal prediction. Initially, JSAM selects meaningful global and regional visual signals to discard irrelevant visual signals and integrates selected visual signals smartly. Subsequently, in a language model, smart JSAM decides at each time-step (level) on how to attend visual or non-visual signals to generate accurate, descriptive, and elegant sentences. Lastly, we examine the efficiency and superiority of the projected model over recent similar image captioning models by conducting essential experimentations on the MS-COCO dataset.

Hide-and-Tell: Learning to Bridge Photo Streams for Visual Storytelling

Visual storytelling is a task of creating a short story based on photo streams. Unlike existing visual captioning, storytelling aims to contain not only factual descriptions, but also human-like narration and semantics. However, the VIST dataset consists only of a small, fixed number of photos per story. Therefore, the main challenge of visual storytelling is to fill in the visual gap between photos with narrative and imaginative story. In this paper, we propose to explicitly learn to imagine a storyline that bridges the visual gap. During training, one or more photos is randomly omitted from the input stack, and we train the network to produce a full plausible story even with missing photo(s). Furthermore, we propose for visual storytelling a hide-and-tell model, which is designed to learn non-local relations across the photo streams and to refine and improve conventional RNN-based models. In experiments, we show that our scheme of hide-and-tell, and the network design are indeed effective at storytelling, and that our model outperforms previous state-of-the-art methods in automatic metrics. Finally, we qualitatively show the learned ability to interpolate storyline over visual gaps.

SibNet: Sibling Convolutional Encoder for Video Captioning.

Visual captioning, the task of describing an image or a video using one or few sentences, is a challenging task owing to the complexity of understanding the copious visual information and describing it using natural language. Motivated by the success of applying neural networks for machine translation, previous work applies sequence to sequence learning to translate videos into sentences. In this work, different from previous work that encodes visual information using a single flow, we introduce a novel Sibling Convolutional Encoder (SibNet) for visual captioning, which employs a dual-branch architecture to collaboratively encode videos. The first content branch encodes visual content information of the video with an autoencoder, capturing the visual appearance information of the video as other networks often do. While the second semantic branch encodes semantic information of the video via visual-semantic joint embedding, which brings complementary representation by considering the semantics when extracting features from videos. Then both branches are effectively combined with soft-attention mechanism and finally fed into a RNN decoder to generate captions. With our SibNet explicitly capturing both content and semantic information, the proposed model can better represent the rich information in videos. To validate the advantages of the proposed model, we conduct experiments on two benchmarks for video captioning, YouTube2Text and MSR-VTT. Our results demonstrate that the proposed SibNet consistently outperforms existing methods across different evaluation metrics.

Survey of deep learning and architectures for visual captioning—transitioning between media and natural languages

Deep Learning Architectures has been researched the most in this decade because of its capability to scale up and solve problems that couldn’t be solved before. Mean while many NLP applications cropped up and there is a requirement to understand how the concepts gradually evolved till date after perceptron was introduced in 1959. This document will provide a detailed description of the computational neuroscience starting from artificial neural network and how researchers retrospected the drawbacks faced by the previous architectures and paved the way for modern deep learning. Modern deep learning is more than what it had been perceived decades ago and had been extended to architectures, with exceptional intelligence, scalability and precision, beyond imagination. This document will provide an overview of the continuation of work and will also specifically deal with applications of various domains related to natural language processing and visual and media contents.

Stateful human-centered visual captioning system to aid video surveillance

The study of Natural Language Generation (NLG), especially how human beings narrate the world, assists in understanding of the visual world for surveillance. Our research proposes an effective technique to axiomatically develop multi-line textual description of visual data by exploiting deep Convolution Neural Networks (CNN). Textual description of visual data aids in providing textual tags for visual information. A human can retrieve elected videos from a repository based on visual tags. Videos contain more complex and detailed information than images and provide more language data. The proposed feats-rich model encodes the visual contents to visual and facial features using CNN architecture. Encoded features are passed to two layer LSTM units with attention mechanism, reducing the number of parameters by encompassing relevant details. Experimental results on Trecvid 2016 and UET-Surveillance dataset depict that model outperforms state-of-the-art methods by scoring BLEU score of 0.35 and 0.52, respectively.

Cross-Modality Retrieval by Joint Correlation Learning

As an indispensable process of cross-media analyzing, comprehending heterogeneous data faces challenges in the fields of visual question answering (VQA), visual captioning, and cross-modality retrieval. Bridging the semantic gap between the two modalities is still difficult. In this article, to address the problem in cross-modality retrieval, we propose a cross-modal learning model with joint correlative calculation learning. First, an auto-encoder is used to embed the visual features by minimizing the error of feature reconstruction and a multi-layer perceptron (MLP) is utilized to model the textual features embedding. Then we design a joint loss function to optimize both the intra- and the inter-correlations among the image-sentence pairs, i.e., the reconstruction loss of visual features, the relevant similarity loss of paired samples, and the triplet relation loss between positive and negative examples. In the proposed method, we optimize the joint loss based on a batch score matrix and utilize all mutual mismatched paired samples to enhance its performance. Our experiments in the retrieval tasks demonstrate the effectiveness of the proposed method. It achieves comparable performance to the state-of-the-art on three benchmarks, i.e., Flickr8k, Flickr30k, and MS-COCO.

Show, Reward, and Tell

Despite the promising progress made in visual captioning and paragraphing, visual storytelling is still largely unexplored. This task is more challenging due to the difficulty in modeling an ordered photo sequence and in generating a relevant paragraph with expressive language style for storytelling. To deal with these challenges, we propose an Attribute-based Hierarchical Generative model with Reinforcement Learning and adversarial training (AHGRL). First, to model the ordered photo sequence and the complex story structure, we propose an attribute-based hierarchical generator. The generator incorporates semantic attributes to create more accurate and relevant descriptions. The hierarchical framework enables the generator to learn from the complex paragraph structure. Second, to generate story-style paragraphs, we design a language-style discriminator, which provides word-level rewards to optimize the generator by policy gradient. Third, we further consider the story generator and the reward critic as adversaries. The generator aims to create indistinguishable paragraphs to human-level stories, whereas the critic aims at distinguishing them and further improving the generator. Extensive experiments on the widely used dataset well demonstrate the advantages of the proposed method over state-of-the-art methods.

Hierarchical LSTMs with Adaptive Attention for Visual Captioning.

Recent progress has been made in using attention based encoder-decoder framework for image and video captioning. Most existing decoders apply the attention mechanism to every generated word including both visual words (e.g., "gun" and "shooting") and non-visual words (e.g., "the", "a"). However, these non-visual words can be easily predicted using natural language model without considering visual signals or attention. Imposing attention mechanism on non-visual words could mislead and decrease the overall performance of visual captioning. Furthermore, the hierarchy of LSTMs enables more complex representation of visual data, capturing information at different scales. Considering these issues, we propose a hierarchical LSTM with adaptive attention (hLSTMat) approach for image and video captioning. Specifically, the proposed framework utilizes the spatial or temporal attention for selecting specific regions or frames to predict the related words, while the adaptive attention is for deciding whether to depend on the visual information or the language context information. Also, a hierarchical LSTMs is designed to simultaneously consider both low-level visual information and high-level language context information to support the caption generation. We design the hLSTMat model as a general framework, and we first instantiate it for the task of video captioning. Then, we further instantiate our hLSTMarefine it and apply it to the imioning task. To demonstrate the effectiveness of our proposed framework, we test our method on both video and image captioning tasks. Experimental results show that our approach achieves the state-of-the-art performance for most of the evaluation metrics on both tasks. The effect of important components is also well exploited in the ablation study.

Deep Learning for Image-to-Text Generation: A Technical Overview

Generating a natural language description from an image is an emerging interdisciplinary problem at the intersection of computer vision, natural language processing, and artificial intelligence (AI). This task, often referred to as image or visual captioning, forms the technical foundation of many important applications, such as semantic visual search, visual intelligence in chatting robots, photo and video sharing in social media, and aid for visually impaired people to perceive surrounding visual content. Thanks to the recent advances in deep learning, the AI research community has witnessed tremendous progress in visual captioning in recent years. In this article, we will first summarize this exciting emerging visual captioning area. We will then analyze the key development and the major progress the community has made, their impact in both research and industry deployment, and what lies ahead in future breakthroughs.

Superimposing visual cues and captions over slides to teach computer operations.

A demonstration of a media package for group instruction in computer operations will be presented. The package consists of slides, captions stored on a filmstrip, and an illustrated workbook. Detailed sequential procedures can be shown to the students through a slide presentation, thus saving expensive computer time. The media packages can also be used by individual students who wish to review the laboratory exercises. Each slide presentation closely follows the steps of a sequential laboratory procedure. This new captioning system allows verbal explanation and cues such as circles and arrows to enhance the meaning of the visuals. The workbook closely follows the slide presentation and includes prints of selected slides. This system allows revision of captions, workbooks, or slides without revision of the entire instructional unit.