ResNet Module Research Articles

A Novel Multi-Feature Fusion Model Based on Pre-Trained Wav2vec 2.0 for Underwater Acoustic Target Recognition

Although recent data-driven Underwater Acoustic Target Recognition (UATR) methods have played a dominant role in marine acoustics, they suffer from complex ocean environments and rather small datasets. To tackle such challenges, researchers have resorted to transfer learning in an effort to fulfill UATR tasks. However, existing pre-trained models are trained on audio speech data, and are not suitable for underwater acoustic data. Therefore, it is necessary to make further optimization on the basis of these models to make them suitable for the UATR task. Here, we propose a novel UATR framework called Attention Layer Supplement Integration (ALSI), which integrates large pre-trained neural networks with customized attention modules for acoustic. Specifically, the ALSI model consists of two important modules, namely Scale ResNet and Residual Hybrid Attention Fusion (RHAF). First, the Scale ResNet module takes the Constant-Q transform feature as input to obtain relatively important frequency information. Next, RHAF takes the temporal feature extracted by wav2vec 2.0 and the frequency feature extracted by Scale ResNet as input and aims to better integrate the time–frequency features with the temporal feature by using the attention mechanism. The RHAF module can help wav2vec 2.0, which is trained on speech data, to better adapt to underwater acoustic data. Finally, the experiments on the ShipsEar dataset demonstrated that our model can achieve recognition accuracy of 96.39%. In conclusion, extensive experiments confirm the effectiveness of our model on the UATR task.

Eye Disease Net: an algorithmic model for rapid diagnosis of diseases.

With the development of science and technology and the improvement of the quality of life, ophthalmic diseases have become one of the major disorders that affect the quality of life of people. In view of this, we propose a new method of ophthalmic disease classification, ED-Net (Eye Disease Classification Net), which is composed of the ED_Resnet model and ED_Xception model, and we compare our ED-Net method with classical classification algorithms, transformer algorithm, more advanced image classification algorithms and ophthalmic disease classification algorithms. We propose the ED_Resnet module and ED_Xception module and reconstruct these two modules into a new image classification algorithm ED-Net, and compared them with classical classification algorithms, transformer algorithms, more advanced image classification algorithms and eye disease classification algorithms.

A Multi-Target Detection Method Based on Improved U-Net for UWB MIMO Through-Wall Radar

Ultra-wideband (UWB) multiple-input multiple-output (MIMO) through-wall radar is widely used in through-wall human target detection for its good penetration characteristics and resolution. However, in actual detection scenarios, weak target masking and adjacent target unresolving will occur in through-wall imaging due to factors such as resolution limitations and differences in human reflectance, which will reduce the probability of target detection. An improved U-Net model is proposed in this paper to improve the detection probability of through-wall targets. In the proposed detection method, a ResNet module and a squeeze-and-excitation (SE) module are integrated in the traditional U-Net model. The ResNet module can reduce the difficulty of feature learning and improve the accuracy of detection. The SE module allows the network to perform feature recalibration and learn to use global information to emphasize useful features selectively and suppress less useful features. The effectiveness of the proposed method is verified via simulations and experiments. Compared with the order statistics constant false alarm rate (OS-CFAR), the fully convolutional networks (FCN) and the traditional U-Net, the proposed method can detect through-wall weak targets and adjacent unresolving targets effectively. The detection precision of the through-wall target is improved, and the missed detection rate is minimized.

Research on Big Data-Driven Urban Traffic Flow Prediction Based on Deep Learning

This paper introduces an innovative approach for the urban traffic flow prediction (TFP) that utilizes big data and deep learning (D-L) to improve accuracy, reducing the incidence of large errors commonplace in traditional methods. By implementing this method, sustainable urban developments are able to be achieved more effectively in the future. First, an Attention-CNN-GRU-ResNet (ACGR) TFP model is built with the D-L network by gridding the urban traffic flow (TF) into a three-dimensional S-T tensor sequence. An attention-based GRU is then introduced to combine spatial and channel attention in the traditional GRU, and the time dependence and spatio-temporal (S-T) heterogeneity of TF in each subset are effectively extracted. Finally, a ResNet module is introduced to capture the S-T dependency, which helps avoid the deep network degradation caused by excessive layers. Results show the proposed method generates the minimum value in RMSE, MAE, and MAPE with 18.32, 10.66, and 5.34, respectively. This research provides a new idea to alleviate data sparsity and consider the difference of input features and offers a novel approach to solve the S-T learning tasks associated with modeling.

Dual Residual Denoising Autoencoder with Channel Attention Mechanism for Modulation of Signals

Aiming to address the problems of the high bit error rate (BER) of demodulation or low classification accuracy of modulation signals with a low signal-to-noise ratio (SNR), we propose a double-residual denoising autoencoder method with a channel attention mechanism, referred to as DRdA-CA, to improve the SNR of modulation signals. The proposed DRdA-CA consists of an encoding module and a decoding module. A squeeze-and-excitation (SE) ResNet module containing one residual connection is modified and then introduced into the autoencoder as the channel attention mechanism, to better extract the characteristics of the modulation signals and reduce the computational complexity of the model. Moreover, the other residual connection is further added inside the encoding and decoding modules to optimize the network degradation problem, which is beneficial for fully exploiting the multi-level features of modulation signals and improving the reconstruction quality of the signal. The ablation experiments prove that both the improved SE module and dual residual connections in the proposed method play an important role in improving the denoising performance. The subsequent experimental results show that the proposed DRdA-CA significantly improves the SNR values of eight modulation types in the range of -12 dB to 8 dB. Especially for 16QAM and 64QAM, the SNR is improved by 8.38 dB and 8.27 dB on average, respectively. Compared to the DnCNN denoising method, the proposed DRdA-CA makes the average classification accuracy increase by 67.59∼74.94% over the entire SNR range. When it comes to the demodulation, compared with the RLS and the DnCNN denoising algorithms, the proposed denoising method reduces the BER of 16QAM by an average of 63.5% and 40.5%, and reduces the BER of 64QAM by an average of 46.7% and 18.6%. The above results show that the proposed DRdA-CA achieves the optimal noise reduction effect.

Real-Time Detection of Mango Based on Improved YOLOv4

Agricultural mechanization occupies a key position in modern agriculture. Aiming at the fruit recognition target detection part of the picking robot, a mango recognition method based on an improved YOLOv4 network structure is proposed, which can quickly and accurately identify and locate mangoes. The method improves the recognition accuracy of the width adjustment network, then reduces the ResNet (Residual Networks) module to adjust the neck network to improve the prediction speed, and finally adds CBAM (Convolutional Block Attention Module) to improve the prediction accuracy of the network. The newly improved network model is YOLOv4-LightC-CBAM. The training results show that the mAP (mean Average Precision) obtained by YOLOV4-LightC-CBAM is 95.12%, which is 3.93% higher than YOLOv4. Regarding detection speed, YOLOV4-LightC-CBAM is up to 45.4 frames, which is 85.3% higher than YOLOv4. The results show that the modified network can recognize mangoes better, faster, and more accurately.

Target Detection Method of UAV Aerial Imagery Based on Improved YOLOv5

Due to the advantages of small size, lightweight, and simple operation, the unmanned aerial vehicle (UAV) has been widely used, and it is also becoming increasingly convenient to capture high-resolution aerial images in a variety of environments. Existing target-detection methods for UAV aerial images lack outstanding performance in the face of challenges such as small targets, dense arrangement, sparse distribution, and a complex background. In response to the above problems, some improvements on the basis of YOLOv5l have been made by us. Specifically, three feature-extraction modules are proposed, using asymmetric convolutions. They are named the Asymmetric ResNet (ASResNet) module, Asymmetric Enhanced Feature Extraction (AEFE) module, and Asymmetric Res2Net (ASRes2Net) module, respectively. According to the respective characteristics of the above three modules, the residual blocks in different positions in the backbone of YOLOv5 were replaced accordingly. An Improved Efficient Channel Attention (IECA) module was added after Focus, and Group Spatial Pyramid Pooling (GSPP) was used to replace the Spatial Pyramid Pooling (SPP) module. In addition, the K-Means++ algorithm was used to obtain more accurate anchor boxes, and the new EIOU-NMS method was used to improve the postprocessing ability of the model. Finally, ablation experiments, comparative experiments, and visualization of results were performed on five datasets, namely CIFAR-10, PASCAL VOC, VEDAI, VisDrone 2019, and Forklift. The effectiveness of the improved strategies and the superiority of the proposed method (YOLO-UAV) were verified. Compared with YOLOv5l, the backbone of the proposed method increased the top-one accuracy of the classification task by 7.20% on the CIFAR-10 dataset. The mean average precision (mAP) of the proposed method on the four object-detection datasets was improved by 5.39%, 5.79%, 4.46%, and 8.90%, respectively.

Dilated convolution based RCNN using feature fusion for Low-Altitude aerial objects

The low-altitude aerial objects are hard to detect by existing deep learning-based object detectors because of the scale variance, small size, and occlusion-related problems. Deep learning-based detectors do not consider contextual information about the scale information of small-sized objects in low-altitude aerial images. This paper proposes a new system using the concept of receptive fields and fusion of feature maps to improve the efficiency of deep object detectors in low-altitude aerial images. A Dilated ResNet Module (DRM) is proposed, motivated from the trident networks, which works on dilated convolutions to study the contextual data for specifically small-sized objects. Applicability of this component builds the model strong towards scale variations in low-altitude aerial objects. Then, Feature Fusion Module (FFM) is created to offer semantic intelligence for better detection of low-altitude aerial objects. We have chosen vastly deployed faster RCNN as the base detector for the proposal of our technique. The dilated convolution-based RCNN using feature fusion (DCRFF) system is implemented on a benchmark low-altitude UAV based-object detection dataset, VisDrone, which contains multiple object categories of pedestrians, vehicles in crowded scenes. The experiments exhibit the enactment of the given detector on chosen low-altitude aerial object dataset. The proposed system of DCRFF achieves 35.04% mAP on the challenging VisDrone dataset, indicating an average improvement of 2% when compared.

PROCEDURE FOR USING NEURAL NETWORKS FOR SEGMENTATION OF RASTER IMAGES

Currently, means of semantic segmentation of images, based on the use of neural networks, are increasingly used in computer systems for various purposes. Despite significant successes in this field, one of the most important unsolved problems is the task of determining the type and parameters of convolutional neural networks, which are the basis of the encoder and decoder. As a result of the research, an appropriate procedure was developed that allows the neural network encoder and decoder to be adapted to the following conditions of the segmentation problem: image size, number of color channels, permissible minimum accuracy of segmentation, permissible maximum computational complexity of segmentation, the need to label segments, the need to select several segments, the need to select deformed, displaced and rotated objects, the maximum computational complexity of learning a neural network model is permissible; admissible training period of the neural network model. The implementation of the procedure of applying neural networks for image segmentation consists in the formation of the basic mathematical support, the construction of the main blocks and the general scheme of the procedure. The developed procedure was verified experimentally on examples of semantic segmentation of images containing objects such as a car. The obtained experimental results show that the application of the proposed procedure allows, avoiding complex long-term experiments, to build a neural network model that, with a sufficiently short training period, ensures the achievement of image segmentation accuracy of about 0.8, which corresponds to the best systems of a similar purpose. It is shown that the ways of further research in the direction of improving the methodological support of neural network segmentation of raster images should be correlated with the justified use of modern modules and mechanisms in the encoder and decoder, adapted to the significant conditions of the given task. For example, the use of the ResNet module allows you to increase the depth of the neural network due to the leveling of the gradient drop effect, and the Inception module provides a reduction in the number of weighting factors and the processing of objects of different sizes.

Human interaction recognition method based on parallel multi-feature fusion network

Human activity recognition is a key technology in intelligent video surveillance and an important research direction in the field of computer vision. However, the complexity of human interaction features and the differences in motion characteristics at different time periods have always existed. In this paper, a human interaction recognition algorithm based on parallel multi-feature fusion network is proposed. First of all, in view of the different amount of information provided by the different time periods of action, an improved time-phased video down sampling method based on Gaussian model is proposed. Second, the Inception module uses different scale convolution kernels for feature extraction. It can improve network performance and reduce the amount of network parameters at the same time. The ResNet module mitigates degradation problem due to increased depth of neural networks and achieves higher classification accuracy. The amount of information provided in the motion video in different stages of motion time is also different. Therefore, we combine the advantages of the Inception network and ResNet to extract feature information, and then we integrate the extracted features. After the extracted features are merged, the training is continued to realize parallel connection of the multi-feature neural network. In this paper, experiments are carried out on the UT dataset. Compared with the traditional activity recognition algorithm, this method can accomplish the recognition tasks of six kinds of interactive actions in a better way, and its accuracy rate reaches 88.9%.

Lemon‐YOLO: An efficient object detection method for lemons in the natural environment

Efficient Intelligent detection is a key technology in automatic harvesting robots. However, citrus detection is still a challenging task because of varying illumination, random occlusion and colour similarity between fruits and leaves in natural conditions. In this paper, a detection method called Lemon-YOLO (L-YOLO) is proposed to improve the accuracy and real-time performance of lemon detection in the natural environment. The SE_ResGNet34 network is designed to replace DarkNet53 network in YOLOv3 algorithm as a new backbone of feature extraction. It can enhance the propagation of features, and needs less parameter, which helps to achieve higher accuracy and speed. Moreover, the SE_ResNet module is added to the detection block, to improve the quality of representations produced from the network by strengthening the convolutional features of channels. The experimental results show that the proposed L-YOLO has an average accuracy(AP) of 96.28% and a detection speed of 106 frames per second (FPS) on the lemon test set, which is 5.68% and 28 FPS higher than the YOLOv3, respectively. The results indicate that the L-YOLO method has superior detection performance. It can recognize and locate lemons in the natural environment more efficiently, providing technical support for the machine's picking lemon and other fruits.

Improved protein model quality assessment by integrating sequential and pairwise features using deep learning.

Accurately estimating protein model quality in the absence of experimental structure is not only important for model evaluation and selection but also useful for model refinement. Progress has been steadily made by introducing new features and algorithms (especially deep neural networks), but the accuracy of quality assessment (QA) is still not very satisfactory, especially local QA on hard protein targets. We propose a new single-model-based QA method ResNetQA for both local and global quality assessment. Our method predicts model quality by integrating sequential and pairwise features using a deep neural network composed of both 1D and 2D convolutional residual neural networks (ResNet). The 2D ResNet module extracts useful information from pairwise features such as model-derived distance maps, co-evolution information, and predicted distance potential from sequences. The 1D ResNet is used to predict local (global) model quality from sequential features and pooled pairwise information generated by 2D ResNet. Tested on the CASP12 and CASP13 datasets, our experimental results show that our method greatly outperforms existing state-of-the-art methods. Our ablation studies indicate that the 2D ResNet module and pairwise features play an important role in improving model quality assessment. https://github.com/AndersJing/ResNetQA. Supplementary data are available at Bioinformatics online.

End-to-End Classification Network for Ice Sheet Subsurface Targets in Radar Imagery

Sea level rise, caused by the accelerated melting of glaciers in Greenland and Antarctica in recent decades, has become a major concern in the scientific, environmental, and political arenas. A comprehensive study of the properties of the ice subsurface targets is particularly important for a reliable analysis of their future evolution. Newer deep learning techniques greatly outperform the traditional techniques based on hand-crafted feature engineering. Therefore, we propose an efficient end-to-end network for the automatic classification of ice sheet subsurface targets in radar imagery. Our network uses bilateral filtering to reduce noise and consists of ResNet module, improved Atrous Spatial Pyramid Pooling (ASPP) module, and decoder module. With radar images provided by the Center of Remote Sensing of Ice Sheets (CReSIS) from 2009 to 2011 as our training and testing data, experimental results confirm the robustness and effectiveness of the proposed network in radargram.

Three-dimensional rapid registration and reconstruction of multi-view rigid objects based on end-to-end deep surface model

Three-dimensional object reconstruction from multi-view images is an important topic in computer vision, which has attracted enormous attention during the past decades. With the further study in deep learning, remarkable progress of three-dimensional object reconstruct has been obtained in recent years. In this paper, we proposed three-dimensional rapid registration and reconstruction of multi-view rigid objects based on end-to-end deep surface model in the field of three-dimensional object reconstruction. Firstly, we introduce a matching algorithm called local stereo matching algorithm based on improved census transform and multi-scale spatial, aiming to improve the matching results for those regions. In cost aggregation step, guided map filtering algorithm with excellent gradient preserving property is introduced into Gaussian pyramid structure and regularization is added to strengthen cost volume consistency. Secondly, the improved inception RESNET module is added to improve the feature extraction ability of the network, and multiple features are extracted by using multiple network structures, and finally multiple features are sequentially input into the VRNN module to enhance the reconstruction effect of multi-view images. The experimental results show that our proposed method can not only achieve better reconstruction results, but also reconstruct more details and spend less time in training.

CASR: a context-aware residual network for single-image super-resolution

With the significant power of deep learning architectures, researchers have made much progress on super-resolution in the past few years. However, due to low representational ability of feature maps extracted from nature scene images, directly applying deep learning architectures for super-resolution could result in poor visual effects. Essentially, unique characteristics like low-frequency information should be emphasized for better shape reconstruction, other than treated equally across different patches and channels. To ease this problem, we propose a lightweight context-aware deep residual network named as CASR network, which appropriately encodes channel and spatial attention information to construct context-aware feature map for single-image super-resolution. We firstly design a task-specified inception block with a novel structure of astrous filters and specially chosen kernel size to extract multi-level information from low-resolution images. Then, a Dual-Attention ResNet module is applied to capture context information by dually connecting spatial and channel attention schemes. With high representational ability of context-aware feature map, CASR can accurately and efficiently generate high-resolution images. Experiments on several popular datasets show the proposed method has achieved better visual improvements and superior efficiencies than most of the existing studies.