Cross-modal Feature Research Articles

Cross-Modal Feature Fusion-Based Knowledge Transfer for Text-Based Person Search

Cross-Modal Feature Fusion-Based Knowledge Transfer for Text-Based Person Search

ECW-EGNet: Exploring Cross-ModalWeighting and edge-guided decoder network for RGB-D salient object detection

Existing RGB-D salient object detection (SOD) techniques concentrate on combining data from multiple modalities (e.g., depth and RGB) and extracting multi-scale data for improved saliency reasoning. However, they frequently perform poorly as a factor of the drawbacks of low-quality depth maps and the lack of correlation between the extracted multi-scale data. In this paper, we propose a Exploring Cross-ModalWeighting and Edge-Guided Decoder Network (ECW-EGNet) for RGB-D SOD, which includes three prominent components. Firstly, we deploy a Cross-Modality Weighting Fusion (CMWF) module that utilizes Channel-Spatial Attention Feature Enhancement (CSAE) mechanism and Depth-Quality Assessment (DQA) mechanism to achieve the cross-modal feature interaction. The former parallels channel attention and spatial attention enhances the features of extracted RGB streams and depth streams while the latter assesses the depth-quality reduces the detrimental influence of the low-quality depth maps during the cross-modal fusion. Then, in order to effectively integrate multi-scale features for high-level and produce salient objects with precise locations, we construct a Bi-directional Scale- Correlation Convolution (BSCC) module in a bi-directional structure. Finally, we construct an Edge-Guided (EG) decoder that uses the edge detection operator to obtain edge masks to guide the enhancement of salient map edge details. The comprehensive experiments on five benchmark RGB-D SOD datasets demonstrate that the proposed ECW-EGNet outperforms 21 state-of-the-art (SOTA) saliency detectors in four widely used evaluation metrics.

MACTFusion: Lightweight Cross Transformer for Adaptive Multimodal Medical Image Fusion.

Multimodal medical image fusion aims to integrate complementary information from different modalities of medical images. Deep learning methods, especially recent vision Transformers, have effectively improved image fusion performance. However, there are limitations for Transformers in image fusion, such as lacks of local feature extraction and cross-modal feature interaction, resulting in insufficient multimodal feature extraction and integration. In addition, the computational cost of Transformers is higher. To address these challenges, in this work, we develop an adaptive cross-modal fusion strategy for unsupervised multimodal medical image fusion. Specifically, we propose a novel lightweight cross Transformer based on cross multi-axis attention mechanism. It includes cross-window attention and cross-grid attention to mine and integrate both local and global interactions of multimodal features. The cross Transformer is further guided by a spatial adaptation fusion module, which allows the model to focus on the most relevant information. Moreover, we design a special feature extraction module that combines multiple gradient residual dense convolutional and Transformer layers to obtain local features from coarse to fine and capture global features. The proposed strategy significantly boosts the fusion performance while minimizing computational costs. Extensive experiments, including clinical brain tumor image fusion, have shown that our model can achieve clearer texture details and better visual quality than other state-of-the-art fusion methods.

GraphCFC: A Directed Graph Based Cross-Modal Feature Complementation Approach for Multimodal Conversational Emotion Recognition

Emotion Recognition in Conversation (ERC) plays a significant part in Human-Computer Interaction (HCI) systems since it can provide empathetic services. Multimodal ERC can mitigate the drawbacks of uni-modal approaches. Recently, Graph Neural Networks (GNNs) have been widely used in a variety of fields due to their superior performance in relation modeling. In multimodal ERC, GNNs are capable of extracting both long-distance contextual information and inter-modal interactive information. Unfortunately, since existing methods such as MMGCN directly fuse multiple modalities, redundant information may be generated and diverse information may be lost. In this work, we present a directed Graph based Cross-modal Feature Complementation (GraphCFC) module that can efficiently model contextual and interactive information. GraphCFC alleviates the problem of heterogeneity gap in multimodal fusion by utilizing multiple subspace extractors and Pair-wise Cross-modal Complementary (PairCC) strategy. We extract various types of edges from the constructed graph for encoding, thus enabling GNNs to extract crucial contextual and interactive information more accurately when performing message passing. Furthermore, we design a GNN structure called GAT-MLP, which can provide a new unified network framework for multimodal learning. The experimental results on two benchmark datasets show that our GraphCFC outperforms the state-of-the-art (SOTA) approaches.

Gloss Prior Guided Visual Feature Learning for Continuous Sign Language Recognition.

Continuous sign language recognition (CSLR) is to recognize the glosses in a sign language video. Enhancing the generalization ability of CSLR's visual feature extractor is a worthy area of investigation. In this paper, we model glosses as priors that help to learn more generalizable visual features. Specifically, the signer-invariant gloss feature is extracted by a pre-trained gloss BERT model. Then we design a gloss prior guidance network (GPGN). It contains a novel parallel densely-connected temporal feature extraction (PDC-TFE) module for multi-resolution visual feature extraction. The PDC-TFE captures the complex temporal patterns of the glosses. The pre-trained gloss feature guides the visual feature learning through a cross-modality matching loss. We propose to formulate the cross-modality feature matching into a regularized optimal transport problem, it can be efficiently solved by a variant of the Sinkhorn algorithm. The GPGN parameters are learned by optimizing a weighted sum of the cross-modality matching loss and CTC loss. The experiment results on German and Chinese sign language benchmarks demonstrate that the proposed GPGN achieves competitive performance. The ablation study verifies the effectiveness of several critical components of the GPGN. Furthermore, the proposed pre-trained gloss BERT model and cross-modality matching can be seamlessly integrated into other RGB-cue-based CSLR methods as plug-and-play formulations to enhance the generalization ability of the visual feature extractor.

Mining Contrastive Relations Between Cross-Modal Features for Zero-Shot Remote Sensing Image Scene Classification

Mining Contrastive Relations Between Cross-Modal Features for Zero-Shot Remote Sensing Image Scene Classification

Cross-Modal Feature Fusion and Interaction Strategy for CNN-Transformer-Based Object Detection in Visual and Infrared Remote Sensing Imagery

Cross-Modal Feature Fusion and Interaction Strategy for CNN-Transformer-Based Object Detection in Visual and Infrared Remote Sensing Imagery

Ultrasound Report Generation with Cross-Modality Feature Alignment via Unsupervised Guidance.

Automatic report generation has arisen as a significant research area in computer-aided diagnosis, aiming to alleviate the burden on clinicians by generating reports automatically based on medical images. In this work, we propose a novel framework for automatic ultrasound report generation, leveraging a combination of unsupervised and supervised learning methods to aid the report generation process. Our framework incorporates unsupervised learning methods to extract potential knowledge from ultrasound text reports, serving as the prior information to guide the model in aligning visual and textual features, thereby addressing the challenge of feature discrepancy. Additionally, we design a global semantic comparison mechanism to enhance the performance of generating more comprehensive and accurate medical reports. To enable the implementation of ultrasound report generation, we constructed three large-scale ultrasound image-text datasets from different organs for training and validation purposes. Extensive evaluations with other state-of-the-art approaches exhibit its superior performance across all three datasets. Code and dataset are valuable at this link.

HealthPrism: A Visual Analytics System for Exploring Children's Physical and Mental Health Profiles with Multimodal Data.

The correlation between children's personal and family characteristics (e.g., demographics and socioeconomic status) and their physical and mental health status has been extensively studied across various research domains, such as public health, medicine, and data science. Such studies can provide insights into the underlying factors affecting children's health and aid in the development of targeted interventions to improve their health outcomes. However, with the availability of multiple data sources, including context data (i.e., the background information of children) and motion data (i.e., sensor data measuring activities of children), new challenges have arisen due to the large-scale, heterogeneous, and multimodal nature of the data. Existing statistical hypothesis-based and learning model-based approaches have been inadequate for comprehensively analyzing the complex correlation between multimodal features and multi-dimensional health outcomes due to the limited information revealed. In this work, we first distill a set of design requirements from multiple levels through conducting a literature review and iteratively interviewing 11 experts from multiple domains (e.g., public health and medicine). Then, we propose HealthPrism, an interactive visual and analytics system for assisting researchers in exploring the importance and influence of various context and motion features on children's health status from multi-levelperspectives. Within HealthPrism, a multimodal learning model with a gate mechanism is proposed for health profiling and cross-modality feature importance comparison. A set of visualization components is designed for experts to explore and understand multimodal data freely. We demonstrate the effectiveness and usability of HealthPrism through quantitative evaluation of the model performance, case studies, and expert interviews in associated domains.

Learning transferable cross-modality representations for few-shot hyperspectral and LiDAR collaborative classification

Hyperspectral image (HSI) classification, incorporating both spatial and spectral information, is a crucial topic in earth observation and land cover analysis. However, ground objects with similar spectral attributes are still the challenges for finer classifications. Recently, deep learning-based multimodality fusion provides promising solutions by exploiting LiDAR data with its geometric information to fuse with spectral attributes. However, the labor-intensive and time-consuming multimodality data annotation limits the performance of supervised deep learning technologies. How to address the semantic disparity between the LiDAR data and HSIs, and learning transferable representations for cross-scene classifications are still challenging. In this paper, we propose a multimodal fusion relational network with meta-learning (MFRN-ML) to solve these challenges. Specifically, the MFRN-ML incorporates the multimodal learning and few-shot learning (FSL) into a three-stage task-based learning framework to learn the transferable cross-modality representations for few-shot HSI and LiDAR collaborative classification. First, a multimodal fusion relational network, composed of a cross-modality feature fusion module and a relation learning module, is proposed to address the challenge of limited annotations in multimodal learning in a data-adaptive way. Then, a three-stage task-based learning framework can train the network to learn transferable representations with few labeled samples for cross-scene classification. We perform experiments on four multimodal datasets collected by different sensors. Compared with existing supervised, semi-supervised, and meta-learning methods, MFRN-ML attains state-of-the-art performances in few-shot tasks. Particularly, our method shows promising generalization ability on unseen categories across different domains.

A survey of multimodal hybrid deep learning for computer vision: Architectures, applications, trends, and challenges

In recent years, deep learning algorithms have rapidly revolutionized artificial intelligence, particularly machine learning, enabling researchers and practitioners to extend previously hand-crafted feature extraction procedures. In particular, deep learning uses adaptive learning processes to learn more complex and informative patterns from datasets of varying sizes. With the increasing availability of multimodal data streams and recent advances in deep learning algorithms, multimodal deep learning is on the rise. This requires the development of complex models that can process and analyze multimodal information in a consistent manner. However, unstructured data can come in many different forms (also known as modalities). Extracting relevant features from this data remains an ambitious goal for deep learning researchers. According to the literature, most deep learning systems consist of a single architecture (i.e., standalone deep learning). When two or more deep learning architectures are combined over multiple sensory modalities, the result is called a multimodal hybrid deep learning model. Since this research direction has received much attention in the field of deep learning, the purpose of this survey is to provide a broader overview of the topic. In this paper, we provide a comprehensive review of recent advances in multimodal hybrid deep learning, including a thorough analysis of the most commonly developed hybrid architectures. In particular, one of the main challenges in multimodal hybrid analysis is the ability of these architectures to systematically integrate cross-modal features in hybrid designs. Therefore, we propose a generic framework for multimodal hybrid learning that focuses mainly on fusion methods. We also identify trends and challenges in multimodal hybrid learning and provide insights and directions for future research. Our findings show that multimodal hybrid learning can perform well in a variety of challenging computer vision applications and tasks.

EFDCNet: Encoding fusion and decoding correction network for RGB-D indoor semantic segmentation

Semantic segmentation is a crucial task in vision measurement systems that involves understanding and segmenting different objects and regions within an image. Over the years, numerous RGB-D semantic segmentation methods have been developed, leveraging the encoder-decoder architecture to achieve outstanding performance. However, existing methods have two main problems that constrain further performance improvement. Firstly, in the encoding stage, existing methods have a weak ability to fuse cross-modal information, and low-quality depth maps can easily lead to poor feature representation. Secondly, in the decoding stage, the upsampling of high-level semantic information may cause the loss of contextual information, and low-level features from the encoder may bring noises to the decoder through skip connections. To solve these issues, we propose a novel Encoding Fusion and Decoding Correction Network (EFDCNet) for RGB-D indoor semantic segmentation. First, in the encoding stage of EFDCNet, we focus on extracting valuable information from low-quality depth maps, and employ a channel-wise filter to select informative depth features. Additionally, we establish the global dependencies between RGB and depth features via the self-attention mechanism to enhance the cross-modal feature interactions, extracting discriminant and powerful features. Then, in the decoding stage of EFDCNet, we use the highest-level information as semantic guidance to compensate for the upsampling information and filter out noise from the low-level encoder features propagated through the skip connections to the decoder. Extensive experiments conducted on two widely-used RGB-D indoor semantic segmentation datasets demonstrate that the proposed EFDCNet surpasses the performance of relevant state-of-the-art methods. The code is available at https://github.com/ Mark9010/EFDCNet

3D Semantic Segmentation of Aerial Photogrammetry Models Based on Orthographic Projection

Semantic segmentation of 3D scenes is one of the most important tasks in the field of computer vision and has attracted much attention. In this paper, we propose a novel framework for 3D semantic segmentation of aerial photogrammetry models, which uses orthographic projection to improve efficiency while still ensuring high precision, and can also be applied to multiple types of models (i.e., textured mesh or colored point cloud). In our pipeline, we first obtain RGB images and elevation images from the 3D scene through orthographic projection, then use the image semantic segmentation network to segment these images to obtain pixel-wise semantic predictions, and finally back-project the segmentation results to the 3D model for fusion. Specifically, for the image semantic segmentation model, we design a cross-modality feature aggregation module and a context guidance module based on category features, which assist the network in learning more discriminative features between different objects. For the 2D-3D semantic fusion, we combine the segmentation results of the 2D images with the geometric consistency of the 3D models for joint optimization, which further improves the accuracy of the 3D semantic segmentation. Extensive experiments on two large-scale urban scenes demonstrate the efficiency and feasibility of our algorithm and surpass the current mainstream 3D deep learning methods.

MMSMCNet: Modal Memory Sharing and Morphological Complementary Networks for RGB-T Urban Scene Semantic Segmentation

Combining color (RGB) images with thermal images can facilitate semantic segmentation of poorly lit urban scenes. However, for RGB-thermal (RGB-T) semantic segmentation, most existing models address cross-modal feature fusion by focusing only on exploring the samples while neglecting the connections between different samples. Additionally, although the importance of boundary, binary, and semantic information is considered in the decoding process, the differences and complementarities between different morphological features are usually neglected. In this paper, we propose a novel RGB-T semantic segmentation network, called MMSMCNet, based on modal memory fusion and morphological multiscale assistance to address the aforementioned problems. For this network, in the encoding part, we used SegFormer for feature extraction of bimodal inputs. Next, our modal memory sharing module implements staged learning and memory sharing of sample information across modal multiscales. Furthermore, we constructed a decoding union unit comprising three decoding units in a layer-by-layer progression that can extract two different morphological features according to the information category and realize the complementary utilization of multiscale cross-modal fusion information. Each unit contains a contour positioning module based on detail information, a skeleton positioning module with deep features as the primary input, and a morphological complementary module for mutual reinforcement of the first two types of information and construction of semantic information. Based on this, we constructed a new supervision strategy, that is, a multi-unit-based complementary supervision strategy. Extensive experiments using two standard datasets showed that MMSMCNet outperformed related state-of-the-art methods. The code is available at: https://github.com/2021nihao/MMSMCNet.

Addressing Information Inequality for Text-Based Person Search via Pedestrian-Centric Visual Denoising and Bias-Aware Alignments

Text-based person search is an important task in video surveillance, which aims to retrieve the corresponding pedestrian images with a given description. In this fine-grained retrieval task, accurate cross-modal information matching is an essential yet challenging problem. However, existing methods usually ignore the information inequality between modalities, which could introduce great difficulties to cross-modal matching. Specifically, in this task, the images inevitably contain some pedestrian-irrelevant noise like background and occlusion, and the descriptions could be biased to partial pedestrian content in images. With that in mind, in this paper, we propose a Text-Guided Denoising and Alignment (TGDA) model to alleviate the information inequality and realize effective cross-modal matching. In TGDA, we first design a prototype-based denoising module, which integrates pedestrian knowledge from textual features into a prototype vector and uses it as guidance to filter out pedestrian-irrelevant noise from visual features. Thereafter, a bias-aware alignment module is introduced, which guides our model to focus on the description-biased pedestrian content in cross-modal features consistently. Through extensive experiments, the effectiveness of both modules has been validated. Besides, our TGDA achieves state-of-the-art performance on various related benchmarks.

Cross Modality Feature Fusion Framework for Diabetic Retinopathy Image Classification

Diabetic retinopathy (DR) is a prevalent ocular condition and a prominent contributor to vision impairment in individuals with diabetes. Consistent monitoring through fundus photography and prompt intervention represents the most efficient strategy for controlling this ailment. Given the substantial diabetic patient population and their extensive screening needs, there is a growing inclination towards harnessing computer-aided and entirely automated methods for diagnosing DR. Over the past years, deep neural networks have achieved remarkable progress across a wide range of applications. Consequently, automating the diagnosis of DR and delivering tailored recommendations to DR patients underscores the significance of accurate and intricate DR classification. In this work, we have present a cross modality feature fusion based framework for diabetic retinopathy (DR) image classification. Here, cross modality means RGB image and it’s green channel. Initially, we have present multi-scale multi- receptive feature extraction block to learn the local and global features from both the modalities. Further, the learned features at various scale are fused effectively with present multi-level feature fusion block for image classification task. We evaluated our present framework on MESSIDOR and IDRID database by comparing it to state-of-the-art (SOTA) deep learning frame- works for DR image classification. The result analysis clearly demonstrate that the present cross-modality feature fusion based classification framework outperforms existing SOTA frameworks in terms of various evaluation parameters.

GMRLNet: A graph-based manifold regularization learning framework for placental insufficiency diagnosis on incomplete multimodal ultrasound data.

Multimodal analysis of placental ultrasound (US) and microflow imaging (MFI) could greatly aid in the early diagnosis and interventional treatment of placental insufficiency (PI), ensuring a normal pregnancy. Existing multimodal analysis methods have weaknesses in multimodal feature representation and modal knowledge definitions and fail on incomplete datasets with unpaired multimodal samples. To address these challenges and efficiently leverage the incomplete multimodal dataset for accurate PI diagnosis, we propose a novel graph-based manifold regularization learning (MRL) framework named GMRLNet. It takes US and MFI images as input and exploits their modality-shared and modality-specific information for optimal multimodal feature representation. Specifically, a graph convolutional-based shared and specific transfer network (GSSTN) is designed to explore intra-modal feature associations, thus decoupling each modal input into interpretable shared and specific spaces. For unimodal knowledge definitions, graph-based manifold knowledge is introduced to describe the sample-level feature representation, local inter-sample relations, and global data distribution of each modality. Then, an MRL paradigm is designed for inter-modal manifold knowledge transfer to obtain effective cross-modal feature representations. Furthermore, MRL transfers the knowledge between both paired and unpaired data for robust learning on incomplete datasets. Experiments were conducted on two clinical datasets to validate the PI classification performance and generalization of GMRLNet. State-of-the-art comparisons show the higher accuracy of GMRLNet on incomplete datasets. Our method achieves 0.913 AUC and 0.904 balanced accuracy (bACC) for paired US and MFI images, as well as 0.906 AUC and 0.888 bACC for unimodal US images, illustrating its application potential in PI CAD systems.

PMG—Pyramidal Multi-Granular Matching for Text-Based Person Re-Identification

Given a textual query, text-based person re-identification is supposed to search for the targeted pedestrian images from a large-scale visual database. Due to the inherent heterogeneity between different modalities, it is challenging to measure the cross-modal affinity between visual and textual data. Existing works typically employ single-granular methods to extract local features and align image regions with relevant words/phrases. Nevertheless, the limited robustness of single-granular methods cannot adapt to the imprecision and variances of visual and textual features, which are usually influenced by the background clutter, position transformation, posture diversity, and occlusion in surveillance videos, thereby leading to the deterioration of cross-modal matching accuracy. In this paper, we propose a Pyramidal Multi-Granular matching network (PMG) that incorporates a gradual transition process between the coarsest global information and the finest local information by a coarse-to-fine pyramidal method for multi-granular cross-modal features extraction and affinities learning. For each body part of a pedestrian, PMG is adequate in ensuring the integrity of local information while minimizing the surrounding interference signals at a certain scale and can adapt to capture discriminative signals of different body parts and achieve semantically alignment between image strips with relevant textual descriptions, thus suppressing the variances of feature extraction and improving the robustness of feature matching. Comprehensive experiments are conducted on the CUHK-PEDES and RSTPReid datasets to validate the effectiveness of the proposed method and results show that PMG outperforms state-of-the-art (SOTA) methods significantly and yields competitive accuracy of cross-modal retrieval.

Swin Transformer-Based Edge Guidance Network for RGB-D Salient Object Detection.

Salient object detection (SOD), which is used to identify the most distinctive object in a given scene, plays an important role in computer vision tasks. Most existing RGB-D SOD methods employ a CNN-based network as the backbone to extract features from RGB and depth images; however, the inherent locality of a CNN-based network limits the performance of CNN-based methods. To tackle this issue, we propose a novel Swin Transformer-based edge guidance network (SwinEGNet) for RGB-D SOD in which the Swin Transformer is employed as a powerful feature extractor to capture the global context. An edge-guided cross-modal interaction module is proposed to effectively enhance and fuse features. In particular, we employed the Swin Transformer as the backbone to extract features from RGB images and depth maps. Then, we introduced the edge extraction module (EEM) to extract edge features and the depth enhancement module (DEM) to enhance depth features. Additionally, a cross-modal interaction module (CIM) was used to integrate cross-modal features from global and local contexts. Finally, we employed a cascaded decoder to refine the prediction map in a coarse-to-fine manner. Extensive experiments demonstrated that our SwinEGNet achieved the best performance on the LFSD, NLPR, DES, and NJU2K datasets and achieved comparable performance on the STEREO dataset compared to 14 state-of-the-art methods. Our model achieved better performance compared to SwinNet, with 88.4% parameters and 77.2% FLOPs. Our code will be publicly available.

AMCFNet: Asymmetric multiscale and crossmodal fusion network for RGB-D semantic segmentation in indoor service robots

Red-green-blue and depth (RGB-D) semantic segmentation is essential for indoor service robots to achieve accurate artificial intelligence. Various RGB-D indoor semantic segmentation methods have been proposed since the widespread adoption of depth maps. These methods have focused mainly on integrating the multiscale and crossmodal features extracted from RGB images and depth maps in the encoder and used unified strategies to recover the local details at the decoder progressively. However, these methods emphasized crossmodal fusion at the encoder, neglecting the distinguishability between RGB and depth features during decoding, thereby undermining the segmentation performance. To adequately exploit the features, we propose an efficient encoder-decoder architecture called asymmetric multiscale and crossmodal fusion network (AMCFNet) endowed with a differential feature integration strategy. Unlike existing methods, we use simple crossmodal fusion at the encoder and design an elaborate decoder to improve the semantic segmentation performance. Specifically, considering high- and low-level features, we propose a semantic aggregation module (SAM) to process the multiscale and crossmodal features in the last three network layers to aggregate high-level semantic information through a cascaded pyramid structure. Moreover, we design a spatial detail supplement module using low-level spatial details from depth maps to adaptively fuse these details and the information obtained from the SAM. Extensive experiments are conducted to demonstrate that the proposed AMCFNet outperforms state-of-the-art approaches.