Adaptive Fusion Network Research Articles

MFVNet: a deep adaptive fusion network with multiple field-of-views for remote sensing image semantic segmentation

MFVNet: a deep adaptive fusion network with multiple field-of-views for remote sensing image semantic segmentation

Radar Emitter Identification with Multi-View Adaptive Fusion Network (MAFN)

Radar emitter identification (REI) aims to extract the fingerprint of an emitter and determine the individual to which it belongs. Although many methods have used deep neural networks (DNNs) for an end-to-end REI, most of them only focus on a single view of signals, such as spectrogram, bi-spectrum, signal waveforms, and so on. When the electromagnetic environment varies, the performance of DNN will be significantly degraded. In this paper, a multi-view adaptive fusion network (MAFN) is proposed by simultaneously exploring the signal waveform and ambiguity function (AF). First, the original waveform and ambiguity function of the radar signals are used separately for feature extraction. Then, a multi-scale feature-level fusion module is constructed for the fusion of multi-view features from waveforms and AF, via the Atrous Spatial Pyramid Pooling (ASPP) structure. Next, the class probability is modeled as Dirichlet distribution to perform adaptive decision-level fusion via evidence theory. Extensive experiments are conducted on two datasets, and the results show that the proposed MAFN can achieve accurate classification of radar emitters and is more robust than its counterparts.

CT synthesis from multi-sequence MRI using adaptive fusion network

ObjectiveTo investigate a method using multi-sequence magnetic resonance imaging (MRI) to synthesize computed tomography (CT) for MRI-only radiation therapy. ApproachWe proposed an adaptive multi-sequence fusion network (AMSF-Net) to exploit both voxel- and context-wise cross-sequence correlations from multiple MRI sequences to synthesize CT using element- and patch-wise fusions, respectively. The element- and patch-wise fusion feature spaces were combined, and the most representative features were selected for modeling. Finally, a densely connected convolutional decoder was applied to utilize the selected features to produce synthetic CT images. Main resultsThis study includes a total number of 90 patients’ T1-weighted MRI, T2-weighted MRI and CT data. The AMSF-Net reduced the average mean absolute error (MAE) from 52.88-57.23 to 49.15 HU, increased the peak signal-to-noise ratio (PSNR) from 24.82-25.32 to 25.63 dB, increased the structural similarity index measure (SSIM) from 0.857-0.869 to 0.878, and increased the dice coefficient of bone from 0.886-0.896 to 0.903 compared to the other three existing multi-sequence learning models. The improvements were statistically significant according to two-tailed paired t-test. In addition, AMSF-Net reduced the intensity difference with real CT in five organs at risk, four types of normal tissue and tumor compared with the baseline models. The MAE decreases in parotid and spinal cord were over 8% and 16% with reference to the mean intensity value of the corresponding organ, respectively. Further, the qualitative evaluations confirmed that AMSF-Net exhibited superior structural image quality of synthesized bone and small organs such as the eye lens. SignificanceThe proposed method can improve the intensity and structural image quality of synthetic CT and has potential for use in clinical applications.

Infrared and Visible Image Fusion via Attention-Based Adaptive Feature Fusion.

Infrared and visible image fusion methods based on feature decomposition are able to generate good fused images. However, most of them employ manually designed simple feature fusion strategies in the reconstruction stage, such as addition or concatenation fusion strategies. These strategies do not pay attention to the relative importance between different features and thus may suffer from issues such as low-contrast, blurring results or information loss. To address this problem, we designed an adaptive fusion network to synthesize decoupled common structural features and distinct modal features under an attention-based adaptive fusion (AAF) strategy. The AAF module adaptively computes different weights assigned to different features according to their relative importance. Moreover, the structural features from different sources are also synthesized under the AAF strategy before reconstruction, to provide a more entire structure information. More important features are thus paid more attention to automatically and advantageous information contained in these features manifests itself more reasonably in the final fused images. Experiments on several datasets demonstrated an obvious improvement of image fusion quality using our method.

Multiscale adaptive fusion network based algorithm for liver tumor detection

Multiscale adaptive fusion network based algorithm for liver tumor detection

A Task-Balanced Multiscale Adaptive Fusion Network for Object Detection in Remote Sensing Images

A Task-Balanced Multiscale Adaptive Fusion Network for Object Detection in Remote Sensing Images

Multiscale Adaptive Fusion Network for Hyperspectral Image Denoising

Removing the noise and improving the visual quality of hyperspectral images (HSIs) is challenging in academia and industry. Great efforts have been made to leverage local, global or spectral context information for HSI denoising. However, existing methods still have limitations in feature interaction exploitation among multiple scales and rich spectral structure preservation. In view of this, we propose a novel solution to investigate the HSI denoising using a Multi-scale Adaptive Fusion Network (MAFNet), which can learn the complex nonlinear mapping between clean and noisy HSI. Two key components contribute to improving the hyperspectral image denoising: A progressively multiscale information aggregation network and a co-attention fusion module. Specifically, we first generate a set of multiscale images and feed them into a coarse-fusion network to exploit the contextual texture correlation. Thereafter, a fine fusion network is followed to exchange the information across the parallel multiscale subnetworks. Furthermore, we design a co-attention fusion module to adaptively emphasize informative features from different scales, and thereby enhance the discriminative learning capability for denoising. Extensive experiments on synthetic and real HSI datasets demonstrate that the proposed MAFNet has achieved better denoising performance than other state-of-the-art techniques. Our codes are available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/summitgao/MAFNet</uri> .

Global Representation Guided Adaptive Fusion Network for Stable Video Crowd Counting

Global Representation Guided Adaptive Fusion Network for Stable Video Crowd Counting

Convolution-Transformer Adaptive Fusion Network for Hyperspectral Image Classification

Hyperspectral image (HSI) classification is an important but challenging topic in the field of remote sensing and earth observation. By coupling the advantages of convolutional neural network (CNN) and Transformer model, the CNN–Transformer hybrid model can extract local and global features simultaneously and has achieved outstanding performance in HSI classification. However, most of the existing CNN–Transformer hybrid models use artificially specified hybrid strategies, which have poor generalization ability and are difficult to meet the requirements of recognizing fine-grained objects in HSI of complex scenes. To overcome this problem, we proposed a convolution–Transformer adaptive fusion network (CTAFNet) for pixel-wise HSI classification. A local–global fusion feature extraction unit, called the convolution–Transformer adaptive fusion kernel, was designed and integrated into the CTAFNet. The kernel captures the local high-frequency features using a convolution module and extracts the global and sequential low-frequency information using a Transformer module. We developed an adaptive feature fusion strategy to fuse the local high-frequency and global low-frequency features to obtain a robust and discriminative representation of the HSI data. An encoder–decoder structure was adopted in the CTAFNet to improve the flow of fused local–global information between different stages, thus ensuring the generalization ability of the model. Experimental results conducted on three large-scale and challenging HSI datasets demonstrate that the proposed network is superior to nine state-of-the-art approaches. We highlighted the effectiveness of adaptive CNN–Transformer hybrid strategy in HSI classification.

Adaptive fusion network for RGB-D salient object detection

Existing state-of-the-art RGB-D saliency detection models mainly utilize the depth information as complementary cues to enhance the RGB information. However, depth maps can be easily influenced by environment and hence are full of noises. Thus, indiscriminately integrating multi-modality (i.e., RGB and depth) features may induce noise-degraded saliency maps. In this paper, we propose a novel Adaptive Fusion Network (AFNet) to solve this problem. Specifically, we design a triplet encoder network consisting of three subnetworks to process RGB, depth, and fused features, respectively. The three subnetworks are interlinked and form a grid net to facilitate mutual refinement of these multi-modality features. Moreover, we propose a Multi-modality Feature Interaction (MFI) module to exploit complementary cues between depth and RGB modalities and adaptively fuse the multi-modality features. Finally, we design the Cascaded Feature Interweaved Decoder (CFID) to exploit complementary information between multi-level features and refine them iteratively to achieve accurate saliency detection. Experimental results on six commonly used benchmark datasets verify that the proposed AFNet outperforms 20 state-of-the-art counterparts in terms of six widely adopted evaluation metrics. Source code will be publicly available athttps://github.com/clelouch/AFNet upon paper acceptance.

Joint strong edge and multi-stream adaptive fusion network for non-uniform image deblurring

Non-uniform motion deblurring has been a challenging problem in the field of computer vision. Currently, deep learning-based deblurring methods have made promising achievements. In this paper, we propose a new joint strong edge and multi-stream adaptive fusion network to achieve non-uniform motion deblurring. The edge map and the blurred map are jointly used as network inputs and Edge Extraction Network (EEN) guides the Deblurring Network (DN) for image recovery and to complement the important edge information. The Multi-stream Adaptive Fusion Module (MAFM) adaptively fuses the edge information and features from the encoder and decoder to reduce feature redundancy to avoid image artifacts. Furthermore, the Dense Attention Feature Extraction Module (DAFEM) is designed to focus on the severely blurred regions of blurry images to obtain important recovery information. In addition, an edge loss function is added to measure the difference of edge features between the generated and clear images to further recover the edges of the deblurred images. Experiments show that our method outperforms currently public methods in terms of PSNR, SSIM and VIF, and generates images with less blur and sharper edges.

MFEAFN: Multi-scale feature enhanced adaptive fusion network for image semantic segmentation.

Low-level features contain spatial detail information, and high-level features contain rich semantic information. Semantic segmentation research focuses on fully acquiring and effectively fusing spatial detail with semantic information. This paper proposes a multiscale feature-enhanced adaptive fusion network named MFEAFN to improve semantic segmentation performance. First, we designed a Double Spatial Pyramid Module named DSPM to extract more high-level semantic information. Second, we designed a Focusing Selective Fusion Module named FSFM to fuse different scales and levels of feature maps. Specifically, the feature maps are enhanced to adaptively fuse these features by generating attention weights through a spatial attention mechanism and a two-dimensional discrete cosine transform, respectively. To validate the effectiveness of FSFM, we designed different fusion modules for comparison and ablation experiments. MFEAFN achieved 82.64% and 78.46% mIoU on the PASCAL VOC2012 and Cityscapes datasets. In addition, our method has better segmentation results than state-of-the-art methods.

Multibranch Adaptive Fusion Network for RGBT Tracking

RGBT tracking has been increasingly investigated in visual tracking due to the strong complementary nature of visible and infrared images. However, in the established RGBT tracking algorithms, multiscale information has not been well exploited and utilized, which limits the performance of the tracker. In this paper, a novel multibranch adaptive fusion network is proposed, which aggregates multiscale information from multiple branches. Specifically, our backbone network draws on the modified VGG-M. To extract the multiscale features, we design a multiscale adapter, which adds two small convolution kernel branches to the backbone in each layer and each modality in a parallel manner. We also design a multibranch fusion module to adaptively aggregate the features from multiple branches and the previous layer. Moreover, we propose a multimodal fusion module for aggregating features between modalities, which could mitigate the impact of noise from low-quality sources. Finally, many results on two recent RGBT tracking datasets show that our method significantly outperforms other state-of-the-art tracking methods.

CCAFNet: Crossflow and Cross-Scale Adaptive Fusion Network for Detecting Salient Objects in RGB-D Images

Owing to the widespread adoption of depth sensors, salient object detection (SOD) supported by depth maps for reliable complementary information is being increasingly investigated. Existing SOD models mainly exploit the relation between an RGB image and its corresponding depth information across three fusion domains: input RGB-D images, extracted feature maps, and output salient object. However, these models do not leverage the crossflows between high- and low-level information well. Moreover, the decoder in these models uses conventional convolution that involves several calculations. To further improve RGB-D SOD, we propose a crossflow and cross-scale adaptive fusion network (CCAFNet) to detect salient objects in RGB-D images. First, a channel fusion module allows for effective fusing depth and high-level RGB features. This module extracts accurate semantic information features from high-level RGB features. Meanwhile, a spatial fusion module combines low-level RGB and depth features with accurate boundaries and subsequently extracts detailed spatial information from low-level depth features. Finally, a purification loss is proposed to precisely learn the boundaries of salient objects and obtain additional details of the objects. The results of comprehensive experiments on seven common RGB-D SOD datasets indicate that the performance of the proposed CCAFNet is comparable to those of state-of-the-art RGB-D SOD models.

AFNet: Adaptive Fusion Network for Remote Sensing Image Semantic Segmentation

Semantic segmentation of remote sensing images plays an important role in many applications. However, a remote sensing image typically comprises a complex and heterogenous urban landscape with objects in various sizes and materials, which causes challenges to the task. In this work, a novel adaptive fusion network (AFNet) is proposed to improve the performance of very high resolution (VHR) remote sensing image segmentation. To coherently label size-varied ground objects from different categories, we design multilevel architecture with the scale-feature attention module (SFAM). By SFAM, at the location of small objects, low-level features from the shallow layers of convolutional neural network (CNN) are enhanced, whilst for large objects, high-level features from deep layers are enhanced. Thus, the features of size-varied objects could be preserved during fusing features from different levels, which helps to label size-varied objects. As for labeling the category with high intra-class difference and varied scales, the multiscale structure with a scale-layer attention module (SLAM) is utilized to learn representative features, where an adjacent score map refinement module (ACSR) is employed as the classifier. By SLAM, when fusing multiscale features, based on the interested objects scale, feature map from appropriate scale is given greater weights. With such a scale-aware strategy, the learned features can be more representative, which is helpful to distinguish objects for semantic segmentation. Besides, the performance is further improved by introducing several nonlinear layers to the ACSR. Extensive experiments conducted on two well-known public high-resolution remote sensing image data sets show the effectiveness of our proposed model. Code and predictions are available at https://github.com/athauna/AFNet/

A multi-path adaptive fusion network for multimodal brain tumor segmentation

The deep learning method has shown its outstanding performance in object recognition and becomes the first choice for medical image analysis. However, how to effectively propagate features in the learning layer and how to fuse low-level visual features and high-level semantic features, is still a challenging task. In addition, with the rapid development of the neural network, an increasing need of running Convolutional Neural Network (CNN) models with limited computing power and memory resource. To address these problems, this paper proposes a novel multi-path adaptive fusion network. More specifically, we apply the idea of “skip connection” in ResNets to the dense block so as to effectively reserve and propagate more low-level visual features. A contiguous memory mechanism has been realized by adopting direction connections from the state of preceding dense block to all layers of current dense block in the network. Then, a multi-path adaptive fusion dense block was adopted in the up-sampling process to adaptively adjust the low-level visual feature and then to fuse with high-level semantic features. By evaluating the proposed framework on the challenging BRATS2015 dataset, it can be proven that this framework achieves state-of-the-art results by comparing with other counterpart methods. Moreover, parameters of the proposed framework are much less than most published methods.

Adaptive and azimuth-aware fusion network of multimodal local features for 3D object detection

This paper focuses on the construction of strong local features and the effective fusion of image and LiDAR data for 3D object detection. We adopt different modalities of LiDAR data to generate rich features and present an adaptive and azimuth-aware network to aggregate local features from image, bird’s eye view maps and point cloud. Our network mainly consists of three subnetworks: ground plane estimation network, region proposal network and adaptive fusion network. The ground plane estimation network extracts features of point cloud and predicts the parameters of a plane which are used for generating abundant 3D anchors. The region proposal network generates features of image and bird’s eye view maps to output region proposals. To integrate heterogeneous image and point cloud features, the adaptive fusion network explicitly adjusts the intensity of multiple local features and achieves the orientation consistency between image and LiDAR data by introducing an azimuth-aware fusion module. Experiments are conducted on KITTI dataset and the results validate the advantages of our aggregation of multimodal local features and the adaptive fusion network.

Multi-Scale and Occlusion Aware Network for Vehicle Detection and Segmentation on UAV Aerial Images

With the advantage of high maneuverability, Unmanned Aerial Vehicles (UAVs) have been widely deployed in vehicle monitoring and controlling. However, processing the images captured by UAV for the extracting vehicle information is hindered by some challenges including arbitrary orientations, huge scale variations and partial occlusion. In seeking to address these challenges, we propose a novel Multi-Scale and Occlusion Aware Network (MSOA-Net) for UAV based vehicle segmentation, which consists of two parts including a Multi-Scale Feature Adaptive Fusion Network (MSFAF-Net) and a Regional Attention based Triple Head Network (RATH-Net). In MSFAF-Net, a self-adaptive feature fusion module is proposed, which can adaptively aggregate hierarchical feature maps from multiple levels to help Feature Pyramid Network (FPN) deal with the scale change of vehicles. The RATH-Net with a self-attention mechanism is proposed to guide the location-sensitive sub-networks to enhance the vehicle of interest and suppress background noise caused by occlusions. In this study, we release a large comprehensive UAV based vehicle segmentation dataset (UVSD), which is the first public dataset for UAV based vehicle detection and segmentation. Experiments are conducted on the challenging UVSD dataset. Experimental results show that the proposed method is efficient in detecting and segmenting vehicles, and outperforms the compared state-of-the-art works.

Light Field Super-Resolution By Jointly Exploiting Internal and External Similarities

Light field images taken by plenoptic cameras often have a tradeoff between spatial and angular resolutions. In this paper, we propose a novel spatial super-resolution approach for light field images by jointly exploiting internal and external similarities. The internal similarity refers to the correlations across the angular dimensions of the 4D light field itself, while the external similarity refers to the cross-scale correlations learned from an external light field dataset. Specifically, we advance the classic projection-based method that exploits the internal similarity by introducing the intensity consistency checking criterion and a back-projection refinement, while the external correlation is learned by a CNN-based method which aggregates all warped high-resolution sub-aperture images upsampled from the low-resolution input using a single image super-resolution method. By analyzing the error distributions of the above two methods and investigating the upperbound of combining them, we find that the internal and external similarities are complementary to each other. Accordingly, we further propose a pixel-wise adaptive fusion network to take advantage of both their merits by learning a weighting matrix. Experimental results on both synthetic and real-world light field datasets validate the superior performance of the proposed approach over the state-of-the-arts.