Search

We consider the problem of referring segmentation in images and videos with natural language. Given an input image (or video) and a referring expression, the goal is to segment the entity referred by the expression in the image or video. In this paper, we propose a cross-modal self-attention (CMSA) module to utilize fine details of individual words and the input image or video, which effectively captures the long-range dependencies between linguistic and visual features. Our model can adaptively focus on informative words in the referring expression and important regions in the visual input. We further propose a gated multi-level fusion (GMLF) module to selectively integrate self-attentive cross-modal features corresponding to different levels of visual features. This module controls the feature fusion of information flow of features at different levels with high-level and low-level semantic information related to different attentive words. Besides, we introduce cross-frame self-attention (CFSA) module to effectively integrate temporal information in consecutive frames which extends our method in the case of referring segmentation in videos. Experiments on benchmark datasets of four referring image datasets and two actor and action video segmentation datasets consistently demonstrate that our proposed approach outperforms existing state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Visual Speech Recognition (VSR) is a task that recognizes speech from external appearances of the face (${\it i}.{\it e}.$ , lips) into text. Since the information from the visual lip movements is not sufficient to fully represent the speech, VSR is considered as one of the challenging problems. One possible way to resolve this problem is additionally utilizing audio which contains rich information for speech recognition. However, the audio information could not be always available such as in crowded situations. Thus, it is necessary to find a way that successfully provides enough information for speech recognition with visual inputs only. In this paper, we alleviate the information insufficiency of visual lip movement by proposing a cross-modal memory augmented VSR with Visual-Audio Memory (VAM). The proposed framework tries to utilize the complementary information of audio even when the audio inputs are not provided at the inference time. Concretely, the proposed VAM learns to imprint audio features of short clip-level into a memory network using the corresponding visual features. To this end, the VAM contains two memories, lip-video key and audio value. We guide the audio value memory to imprint the audio feature and the lip-video key memory to memorize the location of the imprinted audio. By doing this, the VAM can exploit rich audio information by accessing the memory using visual inputs only. Experimental results show that the proposed method achieves state-of-the-art performance on both word- and sentence-level VSR. In addition, we verify the learned representations inside the VAM contain meaningful information for VSR. [ABSTRACT FROM AUTHOR]

With the ever-increasing urbanization process, modeling people's spatiotemporal activities from their online traces has become a crucial task. State-of-the-art methods for this task rely on cross-modal embedding, which maps items from different modalities (e.g., location, time, text) into the same latent space. Despite their inspiring results, existing cross-modal embedding methods merely capture co-occurrences between items without modeling their high-order interactions. In this paper, we first construct two graphs from raw data records to represent the user interaction graph layer and activity graph layer and propose a hierarchical cross-modal embedding method that takes the high-order relationships into consideration. The key notion behind our method is a novel hierarchical embedding framework with meta-graphs connecting different layers. We introduce both inter-record and intra-record meta-graph structures, which enable learning distributed representations that preserve high-order proximities across graphs from different layers. Our empirical experiments on three real-world datasets demonstrate that our method not only outperforms state-of-the-art methods for spatiotemporal activity prediction, but also captures cross-modal proximity at a finer granularity. [ABSTRACT FROM AUTHOR]

Human perceptions of music and image are closely related to each other, since both can inspire similar human sensations, such as emotion, motion, and power. This paper aims to explore whether and how music and image can be automatically matched by machines. The main contributions are three aspects. First, we construct a benchmark dataset composed of more than $45\,000$ music-image pairs. Human labelers are recruited to annotate whether these pairs are well-matched or not. The results show that they generally agree with each other on the matching degree of music-image pairs. Secondly, we investigate suitable semantic representations of music and image for this cross-modal matching task. In particular, we adopt lyrics as a middle-media to connect music and image, and design a set of lyric-based attributes for image representation. Thirdly, we propose cross-modal ranking analysis (CMRA) to learn the semantic similarity between music and image with ranking labeling information. CMRA aims to find the optimal embedding spaces for both music and image in the sense of maximizing the ordinal margin between music–image pairs. The proposed method is able to learn the non-linear relationship between music and image, and to integrate heterogeneous ranking data from different modalities into a unified space. Experimental results demonstrate that the proposed method outperforms state-of-the-art cross-modal methods in the music–image matching task, and achieves a consistency rate of $91.5\%$ with human labelers. [ABSTRACT FROM PUBLISHER]

In this paper, we introduce Recipe1M+, a new large-scale, structured corpus of over one million cooking recipes and 13 million food images. As the largest publicly available collection of recipe data, Recipe1M+ affords the ability to train high-capacity models on aligned, multimodal data. Using these data, we train a neural network to learn a joint embedding of recipes and images that yields impressive results on an image-recipe retrieval task. Moreover, we demonstrate that regularization via the addition of a high-level classification objective both improves retrieval performance to rival that of humans and enables semantic vector arithmetic. We postulate that these embeddings will provide a basis for further exploration of the Recipe1M+ dataset and food and cooking in general. Code, data and models are publicly available. 1. http://im2recipe.csail.mit.edu. [ABSTRACT FROM AUTHOR]

Mobile devices usually mount a depth sensor to resolve ill-posed problems, like salient object detection on cluttered background. The main barrier of exploring RGBD data is to handle the information from two different modalities. To cope with this problem, in this paper, we propose a boundary-aware cross-modal fusion network for RGBD salient object detection. In particular, to enhance the fusion of color and depth features, we present a cross-modal feature sampling module to balance the contribution of the RGB and depth features based on the statistics of their channel values. In addition, in our multi-scale dense fusion network architecture, we not only incorporate edge-sensitive losses to preserve the boundary of the detected salient region, but also refine its structure by merging the estimated saliency maps of different scales. We accomplish the multi-scale saliency map merging using two alternative methods which produce refined saliency maps via per-pixel weighted combination and an encoder-decoder network. Extensive experimental evaluations demonstrate that our proposed framework can achieve the state-of-the-art performance on several public RGBD-based datasets. [ABSTRACT FROM AUTHOR]

This paper proposes a new strategy for learning effective cross-modal joint embeddings using self-supervision. We set up the problem as one of cross-modal retrieval, where the objective is to find the most relevant data in one domain given input in another. The method builds on the recent advances in learning representations from cross-modal self-supervision using contrastive or binary cross-entropy loss functions. To investigate the robustness of the proposed learning strategy across multi-modal applications, we perform experiments for two applications – audio-visual synchronisation and cross-modal biometrics. The audio-visual synchronisation task requires temporal correspondence between modalities to obtain joint representation of phonemes and visemes, and the cross-modal biometrics task requires common speakers representations given their face images and audio tracks. Experiments show that the performance of systems trained using proposed method far exceed that of existing methods on both tasks, whilst allowing significantly faster training. [ABSTRACT FROM AUTHOR]

Multimodal hashing has attracted much interest for cross-modal similarity search on large-scale multimedia data sets because of its efficiency and effectiveness. Recently, supervised multimodal hashing, which tries to preserve the semantic information obtained from the labels of training data, has received considerable attention for its higher search accuracy compared with unsupervised multimodal hashing. Although these algorithms are promising, they are mainly designed to preserve pairwise similarities. When semantic labels of training data are given, the algorithms often transform the labels into pairwise similarities, which gives rise to the following problems: (1) constructing pairwise similarity matrix requires enormous storage space and a large amount of calculation, making these methods unscalable to large-scale data sets; (2) transforming labels into pairwise similarities loses the category information of the training data. Therefore, these methods do not enable the hash codes to preserve the discriminative information reflected by labels and, hence, the retrieval accuracies of these methods are affected. To address these challenges, this paper introduces a simple yet effective supervised multimodal hashing method, called label consistent matrix factorization hashing (LCMFH), which focuses on directly utilizing semantic labels to guide the hashing learning procedure. Considering that relevant data from different modalities have semantic correlations, LCMFH transforms heterogeneous data into latent semantic spaces in which multimodal data from the same category share the same representation. Therefore, hash codes quantified by the obtained representations are consistent with the semantic labels of the original data and, thus, can have more discriminative power for cross-modal similarity search tasks. Thorough experiments on standard databases show that the proposed algorithm outperforms several state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Cross-modal retrieval aims to identify relevant data across different modalities. In this work, we are dedicated to cross-modal retrieval between images and text sentences, which is formulated into similarity measurement for each image-text pair. To this end, we propose a Cross-modal Relation Guided Network (CRGN) to embed image and text into a latent feature space. The CRGN model uses GRU to extract text feature and ResNet model to learn the globally guided image feature. Based on the global feature guiding and sentence generation learning, the relation between image regions can be modeled. The final image embedding is generated by a relation embedding module with an attention mechanism. With the image embeddings and text embeddings, we conduct cross-modal retrieval based on the cosine similarity. The learned embedding space well captures the inherent relevance between image and text. We evaluate our approach with extensive experiments on two public benchmark datasets, i.e., MS-COCO and Flickr30K. Experimental results demonstrate that our approach achieves better or comparable performance with the state-of-the-art methods with notable efficiency. [ABSTRACT FROM AUTHOR]

Early diagnosis and intervention of depression promote complete recovery, with its traditional clinical assessments depending on the diagnostic scales, clinical experience of doctors and patient cooperation. Recent researches indicate that functional near-infrared spectroscopy (fNIRS) based on deep learning provides a promising approach to depression diagnosis. However, collecting large fNIRS datasets within a standard experimental paradigm remains challenging, limiting the applications of deep networks that require more data. To address these challenges, in this paper, we propose an fNIRS-driven depression recognition architecture based on cross-modal data augmentation (fCMDA), which converts fNIRS data into pseudo-sequence activation images. The approach incorporates a time-domain augmentation mechanism, including time warping and time masking, to generate diverse data. Additionally, we design a stimulation task-driven data pseudo-sequence method to map fNIRS data into pseudo-sequence activation images, facilitating the extraction of spatial-temporal, contextual and dynamic characteristics. Ultimately, we construct a depression recognition model based on deep classification networks using the imbalance loss function. Extensive experiments are performed on the two-class depression diagnosis and five-class depression severity recognition, which reveal impressive results with accuracy of 0.905 and 0.889, respectively. The fCMDA architecture provides a novel solution for effective depression recognition with limited data.

Visible and infrared image fusion is an important image enhancement technique that aims to generate high-quality fused images with prominent targets and rich textures in extreme environments. However, most of the current image fusion methods have poor visual perception of the generated fused images due to severe degradation of texture details of the visible light images in scenes with extreme lighting, which seriously affects the application of subsequent advanced vision tasks such as target detection and tracking. To address these challenges, this paper bridges the gap between image fusion and advanced vision tasks by proposing an efficient fusion method for mutual enhancement of spatial and radiometric information. First, we design the gradient residual dense block (LGCnet) to improve the description of fine spatial details in the fusion network. Then, we developed a cross-modal perceptual fusion (CMPF) module to facilitate modal interactions in the network, which effectively enhances the fusion of complementary information between different modalities and reduces redundant learning. Finally, we designed an adaptive light-aware network (ALPnet) to guide the training of the fusion network to facilitate the fusion network to adaptively select more effective information for fusion under different lighting conditions. Extensive experiments show that the proposed fusion approach has competitive advantages over six current state-of-the-art deep-learning methods in highlighting target features and describing the global scene.

Virtual Environments (VEs) provide the opportunity to simulate a wide range of applications, from training to entertainment, in a safe and controlled manner. For applications which require realistic representations of real world environments, the VEs need to provide multiple, physically accurate sensory stimuli. However, simulating all the senses that comprise the human sensory system (HSS) is a task that requires significant computational resources. Since it is intractable to deliver all senses at the highest quality, we propose a resource distribution scheme in order to achieve an optimal perceptual experience within the given computational budgets. This paper investigates resource balancing for multi-modal scenarios composed of aural, visual and olfactory stimuli. Three experimental studies were conducted. The first experiment identified perceptual boundaries for olfactory computation. In the second experiment, participants ($N=25$ ) were asked, across a fixed number of budgets ($M=5$ ), to identify what they perceived to be the best visual, acoustic and olfactory stimulus quality for a given computational budget. Results demonstrate that participants tend to prioritize visual quality compared to other sensory stimuli. However, as the budget size is increased, users prefer a balanced distribution of resources with an increased preference for having smell impulses in the VE. Based on the collected data, a quality prediction model is proposed and its accuracy is validated against previously unused budgets and an untested scenario in a third and final experiment. [ABSTRACT FROM AUTHOR]

Traditional cross-modal hash models enable efficient and fast retrieval between multimodal data by training high-quality hash representations. The key to the cross-modal hashing model is feature extraction. However, the quality of the features largely depends on the semantic similarity between the multi-modal data, and the existing methods do not effectively utilize the semantic information between the data. In this paper, we attempt to explore the semantic information inherent within the data using contrastive learning. Specifically, we propose a end-to-end cluster-level contrastive learning method (SCCDH) for cross-modal hashing. The method utilizes the clustering results to guide feature learning in an appropriately designed contrast framework. In SCCDH, feature-level and hash cluster-level contrastive learning are used to help the model learn discriminative features among multimodal data. In addition, we propose a distillation filtering method to filter out a large amount of noise in the data. Extensive experiments were conducted on the MIRFLICKR-25K, NUS-WIDE, and MS-COCO datasets, and the results demonstrate that the proposed method outperformed several state-of-the-art methods.

Recently, convolutional neural network (CNN) visual features have demonstrated their powerful ability as a universal representation for various recognition tasks. In this paper, cross-modal retrieval with CNN visual features is implemented with several classic methods. Specifically, off-the-shelf CNN visual features are extracted from the CNN model, which is pretrained on ImageNet with more than one million images from 1000 object categories, as a generic image representation to tackle cross-modal retrieval. To further enhance the representational ability of CNN visual features, based on the pretrained CNN model on ImageNet, a fine-tuning step is performed by using the open source Caffe CNN library for each target data set. Besides, we propose a deep semantic matching method to address the cross-modal retrieval problem with respect to samples which are annotated with one or multiple labels. Extensive experiments on five popular publicly available data sets well demonstrate the superiority of CNN visual features for cross-modal retrieval. [ABSTRACT FROM PUBLISHER]

In social media, images and texts are used to convey individuals' attitudes and feelings; thus, social media has become an indispensable part of people's lives. To understand social behavior and provide better recommendations, sentiment analysis on social media is helpful. One sentiment analysis task is polarity prediction. Although current research on visual or textual sentiment analysis has achieved quite good progress, multimodal and cross-modal analysis combining visual and textual correlation is still in the exploration stage. To capture a semantic connection between images and captions, this paper proposes a cross-modal approach that considers both images and captions in classifying image sentiment polarity. This method transfers the correlation between textual content to images. First, the image and its corresponding caption are sent into an inner-class mapping model, where they are transformed into vectors in Hilbert space to obtain their labels by calculating the inner-class maximum mean discrepancy (MMD). Then, a class-aware sentence representation (CASR) model assigns the distributed representation to the labels with a class-aware attention-based gated recurrent unit (GRU). Finally, an inner-class dependency LSTM (IDLSTM) classifies the sentiment polarity. Experiments carried out on the Getty Images dataset and Twitter 1269 dataset demonstrate the effectiveness of our approach. Moreover, extensive experimental results show that our model outperforms baseline solutions.

Cross-modal applications that elaborately integrate audio, video, and haptic streams will become the mainstream of the eHealth systems. However, existing stream schedulers usually fail to simultaneously meet the cross-modal transmission requests in terms of low latency, high reliability, high throughput, and low complexity. To circumvent this dilemma, this article proposes a general cross-modal stream scheduling scheme by fully taking advantage of the characteristics of different modal streams and their underlying temporal, spatial, and semantic relevance. Specifically, we first propose a hierarchical stream category framework, in which the transmission priority of the modal stream instead of the data flow can be flexibly settled. Next, we design a series of modal-aware stream scheduling schemes by jointly making use of the network slice and mobile edge computing to achieve the tradeoff among the various metrics. Importantly, the transmission strategy can be adjusted adaptively to realize the optimal resource allocation. Subsequently, we analyze the relationship among the user experience, multi-modal impact, and stream scheduling through investigating the interacted impacts among the different modal streams, then develop a user experience based scheduling switch strategy to improve the application generality and reduce the performance fluctuation. Numerical objective and subjective results demonstrate the efficiency of the proposed cross-modal scheduling scheme. [ABSTRACT FROM AUTHOR]

The task of text-to-image synthesis is to generate photographic images conditioned on given textual descriptions. This challenging task has recently attracted considerable attention from the multimedia community due to its potential applications. Most of the up-to-date approaches are built based on generative adversarial network (GAN) models, and they synthesize images conditioned on the global linguistic representation. However, the sparsity of the global representation results in training difficulties on GANs and a shortage of fine-grained information in the generated images. To address this problem, we propose cross-modal global and local linguistic representations-based generative adversarial networks (CGL-GAN) by incorporating the local linguistic representation into the GAN. In our CGL-GAN, we construct a generator to synthesize the target images and a discriminator to judge whether the generated images conform with the text description. In the discriminator, we construct the cross-modal correlation by projecting the image representations at high and low levels onto the global and local linguistic representations, respectively. We design the hinge loss function to train our CGL-GAN model. We evaluate the proposed CGL-GAN on two publicly available datasets, the CUB and the MS-COCO. The extensive experiments demonstrate that incorporating fine-grained local linguistic information with cross-modal correlation can greatly improve the performance of text-to-image synthesis, even when generating high-resolution images. [ABSTRACT FROM AUTHOR]

Online product reviews sentiment classification plays an important role on service recommendation, yet most of current researches on it only focus on single-modal information ignoring the complementary information, that results in unsatisfied accuracy of sentiment classification. This paper proposes a cross-modal hypergraph model to capture textual information and sentimental information simultaneously for sentiment classification of reviews. Furthermore, a mixture model by coupling the latent Dirichlet allocation topic model with the proposed cross-modal hypergraph is designed to mitigate the ambiguity of some specific words, which may express opposite polarity in different contexts. Experiments are carried out on four-domain data sets (books, DVD, electronics, and kitchen) to evaluate the proposed approaches by comparison with lexicon-based method, Naïve Bayes, maximum entropy, and support vector machine. Results demonstrate that our schemes outperform the baseline methods in sentiment classification accuracy.

Automatic text summarization is a fundamental natural language processing (NLP) application that aims to condense a source text into a shorter version. The rapid increase in multimedia data transmission over the Internet necessitates multi-modal summarization (MMS) from asynchronous collections of text, image, audio, and video. In this work, we propose an extractive MMS method that unites the techniques of NLP, speech processing, and computer vision to explore the rich information contained in multi-modal data and to improve the quality of multimedia news summarization. The key idea is to bridge the semantic gaps between multi-modal content. Audio and visual are main modalities in the video. For audio information, we design an approach to selectively use its transcription and to infer the salience of the transcription with audio signals. For visual information, we learn the joint representations of text and images using a neural network. Then, we capture the coverage of the generated summary for important visual information through text-image matching or multi-modal topic modeling. Finally, all the multi-modal aspects are considered to generate a textual summary by maximizing the salience, non-redundancy, readability, and coverage through the budgeted optimization of submodular functions. We further introduce a publicly available MMS corpus in English and Chinese.1 The experimental results obtained on our dataset demonstrate that our methods based on image matching and image topic framework outperform other competitive baseline methods. [ABSTRACT FROM AUTHOR]