Low-resolution Features Research Articles

UN-YOLOv5s: A UAV-Based Aerial Photography Detection Algorithm.

With the progress of science and technology, artificial intelligence is widely used in various disciplines and has produced amazing results. The research of the target detection algorithm has significantly improved the performance and role of unmanned aerial vehicles (UAVs), and plays an irreplaceable role in preventing forest fires, evacuating crowded people, surveying and rescuing explorers. At this stage, the target detection algorithm deployed in UAVs has been applied to production and life, but making the detection accuracy higher and better adaptability is still the motivation for researchers to continue to study. In aerial images, due to the high shooting height, small size, low resolution and few features, it is difficult to be detected by conventional target detection algorithms. In this paper, the UN-YOLOv5s algorithm can solve the difficult problem of small target detection excellently. The more accurate small target detection (MASD) mechanism is used to greatly improve the detection accuracy of small and medium targets, The multi-scale feature fusion (MCF) path is combined to fuse the semantic information and location information of the image to improve the expression ability of the novel model. The new convolution SimAM residual (CSR) module is introduced to make the network more stable and focused. On the VisDrone dataset, the mean average precision (mAP) of UAV necessity you only look once v5s(UN-YOLOv5s) is 8.4% higher than that of the original algorithm. Compared with the same version, YOLOv5l, the mAP is increased by 2.2%, and the Giga Floating-point Operations Per Second (GFLOPs) is reduced by 65.3%. Compared with the same series of YOLOv3, the mAP is increased by 1.8%, and GFLOPs is reduced by 75.8%. Compared with the same series of YOLOv8s, the detection accuracy of the mAP is improved by 1.1%.

Selector-Enhancer: Learning Dynamic Selection of Local and Non-local Attention Operation for Speech Enhancement

Attention mechanisms, such as local and non-local attention, play a fundamental role in recent deep learning based speech enhancement (SE) systems. However, a natural speech contains many fast-changing and relatively briefly acoustic events, therefore, capturing the most informative speech features by indiscriminately using local and non-local attention is challenged. We observe that the noise type and speech feature vary within a sequence of speech and the local and non-local can respectively process different types of corrupted speech regions. To leverage this, we propose Selector-Enhancer, a dual-attention based convolution neural network (CNN) with a feature-filter that can dynamically select regions from low-resolution speech features and feed them to local or non-local attention operations. In particular, the proposed feature-filter is trained by using reinforcement learning (RL) with a developed difficulty-regulated reward that related to network performance, model complexity and “the difficulty of the SE task”. The results show that our method achieves comparable or superior performance to existing approaches. In particular, Selector-Enhancer is effective for real-world denoising, where the number and types of noise are varies on a single noisy mixture.

Hyperspectral Super-Resolution Reconstruction Network Based on Hybrid Convolution and Spectral Symmetry Preservation

Hyperspectral images (HSI) have high-dimensional and complex spectral characteristics, with dozens or even hundreds of bands covering the same area of pixels. The rich information of the ground objects makes hyperspectral images widely used in satellite remote sensing. Due to the limitations of remote sensing satellite sensors, hyperspectral images suffer from insufficient spatial resolution. Therefore, utilizing software algorithms to improve the spatial resolution of hyperspectral images has become an urgent problem that needs to be solved. The spatial information and spectral information of hyperspectral images are strongly correlated. If only the spatial resolution is improved, it often damages the spectral information. Inspired by the high correlation between spectral information in adjacent spectral bands of hyperspectral images, a hybrid convolution and spectral symmetry preservation network has been proposed for hyperspectral super-resolution reconstruction. This includes a model to integrate information from neighboring spectral bands to supplement target band feature information. The proposed model introduces flexible spatial-spectral symmetric 3D convolution in the network structure to extract low-resolution and neighboring band features. At the same time, a combination of deformable convolution and attention mechanisms is used to extract information from low-resolution bands. Finally, multiple bands are fused in the reconstruction module, and the high-resolution hyperspectral image containing global information is obtained by Fourier transform upsampling. Experiments were conducted on the indoor hyperspectral image dataset CAVE, the airborne hyperspectral dataset Pavia Center, and Chikusei. In the X2 super-resolution task, the PSNR values achieved on the CAVE, Pavia Center, and Chikusei datasets were 46.335, 36.321, and 46.310, respectively. In the X4 super-resolution task, the PSNR values achieved on the CAVE, Pavia Center, and Chikusei datasets were 41.218, 30.377, and 38.365, respectively. The results show that our method outperforms many advanced algorithms in objective indicators such as PSNR and SSIM while maintaining the spectral characteristics of hyperspectral images.

Multi-object tracking via deep feature fusion and association analysis

We describe a tracking-by-detection framework for multi-object tracking (MOT). It first detects the objects of interest in each frame of the video, followed by identifying association with the object detected in the previous frame. A deep association network is described to perform object feature matching in the arbitrary two frames to infer association degree of objects, and then similarity matrix loss is used to calculate association between each object in different frames to achieve an accurate tracking. The novelty of the work lies in the design of a multi-scale fusion strategy by gradually concatenating sub-networks of low-resolution feature maps in parallel to the main network of high-resolution feature maps, in the construction of a deeper backbone network which can enhance the semantic information of object features, and in the use of a siamese network for training a pair of discontinuous frames. The main advantage of our framework is that it avoids missing detection and partial detection. It is particularly suitable for solving the problem of object ID switch caused by occlusion, entering and leaving of objects. Our method is evaluated and demonstrated on the publicly available MOT15, MOT16, MOT17 and MOT20 benchmark datasets. Compared with the state-of-the-art methods, our method achieves better tracking performance, and is therefore, more suited for MOT tasks.

Localization of Inpainting Forgery With Feature Enhancement Network

Inpainting the given region of an image is a typical requirement in computer vision. Conventional inpainting, through exemplar-based or diffusion-based strategies, can create realistic inpainted images at a very low cost. Also, such easy-to-use manipulation poses new security threats. Therefore, the detection of inpainting has attracted considerable attention from researchers. However, the existing methods are typically not suitable for the general detection of various inpainting algorithms. Motivated by this, in this work, an efficient feature enhancement network is proposed to locate the inpainted regions in the digital image. Firstly, we design an artifact enhancement block to effectively capture the traces left by diffusion or exemplar-based inpainting. Then, the VGGNet is used as a feature extractor to describe advanced and low-resolution features. Finally, to take full advantage of enhanced features, we concatenate the features obtained by the feature extractor and the up-sampling operations. Extensive experimental evaluations, covering benchmarking, ablation, robustness, generalization, and efficiency studies, confirm the usefulness of the proposed method. This is especially true on the conventional inpainting dataset, our method obtains an average F1 score 7.63% higher than the second-best method. Theoretical and numerical analyses support the effectiveness of our feature enhancement network in representing the artifacts in inpainted images, exhibiting better potential for real-world forensics than various state-of-the-art strategies.

SADENet: A supervised attention delicate enhanced network for subtle person detection

Object detection algorithms have been developed for decades, and one of remaining important challenges is the detection of subtle targets in complex environments. Many convolution neural network (CNN) based methods have been proposed to handle this problem. However, the information pertaining to subtle targets is delicate and can easily be lost or neglected in the upsampling and downsampling stages of conventional CNN. In this paper, we propose a new subtle target detection network, named the supervised attention delicate enhanced network (SADENet). Firstly, two sample vertebras are designed to solve the problem of information loss during upsampling and downsampling. Then, an enhanced context attention module (ECAM) is proposed, which extends the original supervised heatmap by fusing information from different scales. Based on the ECAM and the respective advantages of high-resolution and low-resolution feature maps, two effective sampling attention modules, the semantic fidelity upsampling module and the location enhancement downsampling module, are created to build a bi-directional connection neck structure. The effectiveness of our proposed method is demonstrated in experiments on the TinyPerson dataset, which outperforms many state-of-the-art detectors, especially in the detection of tiny and tiny1 scale targets.

SSP-Net: Scalable sequential pyramid networks for real-Time 3D human pose regression

In this paper we propose a highly scalable convolutional neural networks, end-to-end trainable, for real-time 3D human pose regression from still RGB images. We call this approach Scalable Sequential Pyramid Networks (SSP-Net) as it is trained with refined supervision at multiple scales in a sequential manner. Our network requires a single training procedure and is capable of producing its best predictions at 120 frames per second (FPS), or acceptable predictions at more than 200 FPS when cut at test time. We show that the proposed regression approach is invariant to the size of feature maps, allowing our method to perform multi-resolution intermediate supervisions and reaching results comparable to the state-of-the-art with very low resolution feature maps. We demonstrate the accuracy and the effectiveness of our method by providing extensive experiments on two of the most important publicly available datasets for 3D pose estimation, Human3.6M and MPI-INF-3DHP. Additionally, we provide relevant insights about our decisions on the network architecture and show its flexibility to meet the best precision-speed compromise.

CFRWD-GAN for SAR-to-Optical Image Translation

Synthetic aperture radar (SAR) images have been extensively used in earthquake monitoring, resource survey, agricultural forecasting, etc. However, it is a challenge to interpret SAR images with severe speckle noise and geometric deformation due to the nature of radar imaging. The translation of SAR-to-optical images provides new support for the interpretation of SAR images. Most of the existing translation networks, which are based on generative adversarial networks (GANs), are vulnerable to part information loss during the feature reasoning stage, making the outline of the translated images blurred and semantic information missing. Aiming to solve these problems, cross-fusion reasoning and wavelet decomposition GAN (CFRWD-GAN) is proposed to preserve structural details and enhance high-frequency band information. Specifically, the cross-fusion reasoning (CFR) structure is proposed to preserve high-resolution, detailed features and low-resolution semantic features in the whole process of feature reasoning. Moreover, the discrete wavelet decomposition (WD) method is adopted to handle the speckle noise in SAR images and achieve the translation of high-frequency components. Finally, the WD branch is integrated with the CFR branch through an adaptive parameter learning method to translate SAR images to optical ones. Extensive experiments conducted on two publicly available datasets, QXS-SAROPT and SEN1-2, demonstrate a better translation performance of the proposed CFRWD-GAN compared to five other state-of-the-art models.

Image super-resolution with multi-scale fractal residual attention network

Deep neural networks can significantly improve the quality of super-resolution. However, previous work has made insufficient use of low-resolution scale features and channel-wise information, hence hindering the representational ability of CNNs. To address these issues, a multi-scale fractal residual attention network (MFRAN) is proposed. Specifically, MFRAN consists of fractal residual blocks (FRBs), dual-enhanced channel attention (DECA), and dilated residual attention blocks (DRABs). Among them, FRB applies multi-scale extension rule to continuously expand into a fractal structure that detects multi-scale features; DRAB constructs a combined dilated convolution to learn a generalizable and expressive feature space with a larger receptive field; DECA employs one-dimensional convolution to achieve cross-channel information interaction, and enhance the flow of information between groups by channel shuffling. Then, we integrate horizontal feature representations via local residual and feature fusion. Extensive quantitative and qualitative evaluations of benchmark datasets show that our proposed approach outperforms state-of-the-art methods in terms of quantitative metrics and visual results.

Super-resolution reconstruction for the three-dimensional turbulence flows with a back-projection network

Recent attempts to employ deep learning technology for the super-resolution (SR) reconstruction of turbulence have focused chiefly on reconstructing two-dimensional (2D) slices of the three-dimensional (3D) flow fields. However, tomographic particle image velocimetry (Tomo-PIV) measurements yield 3D velocity fields. The resolutions of 3D velocity field data along three directions cannot be simultaneously enhanced using a 2D SR reconstruction model. Additionally, a 3D velocity flow field structure cannot be accurately reconstructed using a 2D model. Therefore, we present a 3D flow field SR reconstruction method (FSR-3D) based on a back-projection network. The FSR-3D model incorporates a multiscale convolutional residual block and an up-and-down projection module and establishes the mapping relationship between low-resolution features and high-resolution (HR) features through an iterative mechanism. The proposed model is validated using two experiments. First, the original velocity field is reconstructed from downsampled forced isotropic turbulence data. Second, the downsampled direct numerical simulation (DNS) data of a turbulent channel flow are used to reconstruct the HR flow field to verify the reconstruction capability of the FSR-3D model with respect to the flow field of the boundary layer. Third, Tomo-PIV measurements of the wake flow behind a circular cylinder are used to further verify the generalization ability of the model and the reconstruction effect of the measured data. The experimental results demonstrate that the proposed FSR-3D model can accurately reconstruct the HR 3D velocity field. At the same time, the properties of the reconstruction results, such as their vortex structure, kinetic energy spectrum, and Reynolds stress, are closer to the DNS results than those of the 2D model. The reconstruction results of the Tomo-PIV measurement data show that the FSR-3D model has good generalization ability.

From degrade to upgrade: Learning a self-supervised degradation guided adaptive network for blind remote sensing image super-resolution

Over the past few years, single image super-resolution (SR) has become a hotspot in the remote sensing area, and numerous methods have made remarkable progress in this fundamental task. However, they usually rely on the assumption that images suffer from a fixed known degradation process, e.g., bicubic downsampling. To save us from performance drop when real-world distribution deviates from the naive assumption, blind image super-resolution for multiple and unknown degradations has been explored. Nevertheless, the lack of a real-world dataset and the challenge of reasonable degradation estimation hinder us from moving forward. In this paper, a self-supervised degradation-guided adaptive network is proposed to mitigate the domain gap between simulation and reality. Firstly, the complicated degradations are characterized by robust representations in embedding space, which promote adaptability to the downstream SR network with degradation priors. Specifically, we incorporated contrastive learning to blind remote sensing image SR, which guides the reconstruction process by encouraging the positive representations (relevant information) while punishing the negatives. Besides, an effective dual-wise feature modulation network is proposed for feature adaptation. With the guide of degradation representations, we conduct modulation on feature and channel dimensions to transform the low-resolution features into the desired domain that is suitable for reconstructing high-resolution images. Extensive experiments on three mainstream datasets have demonstrated our superiority against state-of-the-art methods. Our source code can be found at https://github.com/XY-boy/DRSR

(MLE$^{2}$A$^{2}$U)-Net: Image Super-Resolution via Multi-Level Edge Embedding and Aggregated Attentive Upsampler Network

Given a degraded low-resolution input image, super-resolution (SR) aims at restoring the lost textures and structures and generating high-resolution image content. Significant advances in image super-resolution have been made lately, dominated by convolutional neural networks (CNNs). The top performing CNN-based SR networks typically employ very deep models for embracing the benefits of generating spatially precise results, but at the cost of loss of long-term contextual information. Additionally, state-of-the-art (SOTA) methods generally lack in maintaining the balance between spatial details and contextual information, which is the basic requirement for exhibiting superior performance in SR task. For restoration application like SR, the overall network generally demands efficient preservation of low-frequency information and reconstruction of high-frequency details. Thus, our work presents a novel architecture with the holistic objective of maintaining spatially-precise representation by collecting contextual content and restoring multi-frequency information throughout the network. Our proposed model learns an enriched set of features, that besides combining contextual information from multiple scales simultaneously preserves the high-resolution spatial details. The core of our approach is a novel non-local and local attention (NLLA) block which focuses on (1) learning enriched features by collecting information from multiple scales, (2) simultaneously handling the different frequency information, and (3) effectively fusing the relevant low-frequency and high-frequency information by ignoring the redundant features. Additionally, for effectively mapping the low-resolution features to high-resolution, we propose a novel aggregated attentive up-sampler (AAU) block that attentively learns the weights to up-sample the refined low-resolution feature maps to high-resolution output. Extensive experiments on the benchmark SR datasets demonstrate that the proposed method achieves appealing performance, both qualitatively and quantitatively.

SR-FEINR: Continuous Remote Sensing Image Super-Resolution Using Feature-Enhanced Implicit Neural Representation

Remote sensing images often have limited resolution, which can hinder their effectiveness in various applications. Super-resolution techniques can enhance the resolution of remote sensing images, and arbitrary resolution super-resolution techniques provide additional flexibility in choosing appropriate image resolutions for different tasks. However, for subsequent processing, such as detection and classification, the resolution of the input image may vary greatly for different methods. In this paper, we propose a method for continuous remote sensing image super-resolution using feature-enhanced implicit neural representation (SR-FEINR). Continuous remote sensing image super-resolution means users can scale a low-resolution image into an image with arbitrary resolution. Our algorithm is composed of three main components: a low-resolution image feature extraction module, a positional encoding module, and a feature-enhanced multi-layer perceptron module. We are the first to apply implicit neural representation in a continuous remote sensing image super-resolution task. Through extensive experiments on two popular remote sensing image datasets, we have shown that our SR-FEINR outperforms the state-of-the-art algorithms in terms of accuracy. Our algorithm showed an average improvement of 0.05 dB over the existing method on ×30 across three datasets.

Simulation of unsteady flow around bluff bodies using knowledge-enhanced convolutional neural network

The unsteady flow with massive separation poses challenges to accurately and efficiently simulate wind effects on civil structures, especially in the search for optimal aerodynamic shapes during the preliminary design stage. To this end, an encoding-transformation-decoding convolutional neural network (CNN) architecture is developed in this study to take the pixelated geometry of a bluff body as the input and instantly output the corresponding unsteady aerodynamic flow (represented by a sequence of consecutive flow snapshots). First, a unique (shared) encoding part is used to extract the low-resolution features from the structural shapes and associated initial flow conditions. Then, the transformation part consecutively maps the encoded features (originated from encoder) to a prediction of next flow snapshot in high-level feature space while the separate decoding part unravels the abstract features (from corresponding transformer) to pixelated velocity and pressure fields. Prior domain knowledge is leveraged at various stages of machine learning pipeline to enhance the learning efficiency and accuracy of the proposed convolutional encoder-transformer-decoder. Specifically, the qualitative knowledge that the flow fields vary on a very small length scale within the boundary layer is used to empirically design a nonuniform grid for CNN image inputs and outputs. In addition, the equation-free quantitative knowledge learned from previous flow snapshot (i.e., high-level features from the transformer and filters in the decoder) is employed in the learning of current flow snapshot as the CNN initial conditions based on transfer learning technique. Furthermore, the equation-based quantitative knowledge of both mass and momentum conservation laws is utilized to augment the CNN loss function. Numerical examples suggest good performance of the proposed knowledge-enhanced CNN for unsteady flow simulation of bridge decks. Effects of these three types of knowledge on the performance of the convolutional encoder-transformer-decoder are also examined to highlight their high efficacy to facilitate the training efficiency and simulation accuracy.

High-Resolution Swin Transformer for Automatic Medical Image Segmentation.

The resolution of feature maps is a critical factor for accurate medical image segmentation. Most of the existing Transformer-based networks for medical image segmentation adopt a U-Net-like architecture, which contains an encoder that converts the high-resolution input image into low-resolution feature maps using a sequence of Transformer blocks and a decoder that gradually generates high-resolution representations from low-resolution feature maps. However, the procedure of recovering high-resolution representations from low-resolution representations may harm the spatial precision of the generated segmentation masks. Unlike previous studies, in this study, we utilized the high-resolution network (HRNet) design style by replacing the convolutional layers with Transformer blocks, continuously exchanging feature map information with different resolutions generated by the Transformer blocks. The proposed Transformer-based network is named the high-resolution Swin Transformer network (HRSTNet). Extensive experiments demonstrated that the HRSTNet can achieve performance comparable with that of the state-of-the-art Transformer-based U-Net-like architecture on the 2021 Brain Tumor Segmentation dataset, the Medical Segmentation Decathlon's liver dataset, and the BTCV multi-organ segmentation dataset.

A facial geometry based detection model for face manipulation using CNN-LSTM architecture

This issue of DeepFake technique that may cause great threat to privacy, democracy, and national security has attracted the attention of deep learning researchers. DeepFake detection, therefore, has been a very hot issue in deep learning research. The face landmark feature maps are often used by many DeepFake approaches in generating fake faces. It also provides key information to help to detect manipulated face images. In this paper, we propose a detection approach for manipulated face images. To make full use of face landmark information, we propose a facial geometry prior module (FGPM) to extract facial geometry feature maps. Then the facial geometry feature maps are embedded into upsampling feature maps generated by the CNN-LSTM network. The proposed FGPM exploits facial maps and frequency domain correlation to analyze the discriminative characteristics between manipulated and non-manipulated regions by incorporating the CNN-LSTM network. Finally, a decoder is used to learn the mapping from low-resolution feature maps to pixel-wise to predict manipulation localization. Or a softmax classifier is used to predict real or fake face images. By experimenting on several popular datasets, the proposed detection model has demonstrated the capability of localizing manipulation at the pixel level and with a high prediction on real or fake face images.

A Study on the Super Resolution Combining Spatial Attention and Channel Attention

Existing CNN-based super resolution methods have low emphasis on high-frequency features, resulting in poor performance for contours and textures. To solve this problem, this paper proposes single image super resolution using an attention mechanism that emphasizes high-frequency features and a feature extraction process with different depths. In order to emphasize the high-frequency features of the channel and space, it is composed of CSBlock that combines channel attention and spatial attention. Attention block using 10 CSBlocks was used for high-frequency feature extraction. In order to extract various features with different degrees of feature emphasis from insufficient low-resolution features, features were extracted from structures connected with different numbers of attention blocks. The extracted features were expanded through sub-pixel convolution to create super resolution images, and learning was performed through L1 loss. Compared to the existing deep learning method, it showed improved results in several high-frequency features such as small object outlines and line patterns. In PSNR and SSIM, it showed about 11% to 26% improvement over the existing Bicubic interpolation and about 1 to 2% improvement over VDSR and EDSR.

Deep Dual-Resolution Networks for Real-Time and Accurate Semantic Segmentation of Traffic Scenes

Using light-weight architectures or reasoning on low-resolution images, recent methods realize very fast scene parsing, even running at more than 100 FPS on a single GPU. However, there is still a significant gap in performance between these real-time methods and the models based on dilation backbones. To this end, we proposed a family of deep dual-resolution networks (DDRNets) for real-time and accurate semantic segmentation, which consist of deep dual-resolution backbones and enhanced low-resolution contextual information extractors. The two deep branches and multiple bilateral fusions of backbones generate higher quality details compared to existing two-pathway methods. The enhanced contextual information extractor named Deep Aggregation Pyramid Pooling Module (DAPPM) enlarges effective receptive fields and fuses multi-scale context based on low-resolution feature maps with little time cost. Our method achieves a new state-of-the-art trade-off between accuracy and speed on both Cityscapes and CamVid dataset. For the input of full resolution, on a single 2080Ti GPU without hardware acceleration, DDRNet-23-slim yields 77.4% mIoU at 102 FPS on Cityscapes test set and 74.7% mIoU at 230 FPS on CamVid test set. With widely used test augmentation, our method is superior to most state-of-the-art models and requires much less computation. Codes and trained models are available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/ydhongHIT/DDRNet</uri> .

Progressive Transformer Machine for Natural Character Reenactment

Character reenactment aims to control a target person’s full-head movement by a driving monocular sequence that is made up of the driving character video. Current algorithms utilize convolution neural networks in generative adversarial networks, which extract historical and geometric information to iteratively generate video frames. However, convolution neural networks can merely capture local information with limited receptive fields and ignore global dependencies that play a crucial role in face synthesis, leading to generating unnatural video frames. In this work, we design a progressive transformer module that introduces multi-head self-attention with convolution refinement to simultaneously capture global-local dependencies. Specifically, we utilize the non-lapping window-based multi-head self-attention mechanism with hierarchical architecture to obtain the larger receptive fields at low-resolution feature map and thus extract global information. To better model local dependencies, we introduce the convolution operation to further refine the attentional weight in the multi-head self-attention mechanism. Finally, we use several stacked progressive transformer modules with the down-sampling operation to encode information of appearance information of previously generated frames and parameterized 3D face information of the current frame. Similarly, we use several stacked progressive transformer modules with the up-sampling operation to iteratively generate video frames. In this way, it can capture global-local information to facilitate generating video frames that are globally natural while preserving sharp outlines and rich detail information. Extensive experiments on several standard benchmarks suggest that the proposed method outperforms current leading algorithms.

Real-Time Moving Ship Detection from Low-Resolution Large-Scale Remote Sensing Image Sequence

Optical remote sensing ship target detection has become an essential means of ocean supervision, coastal defense, and frontier defense. Accurate, effective, fast, and real-time remote sensing data processing is the critical technology in this field. This paper proposes a real-time detection algorithm for moving targets in low-resolution wide-area remote sensing images, which includes four steps: pre-screening, simplified HOG feature identification, sequence correlation identification, and facilitated Yolo identification. It can effectively detect and track targets in low-resolution sequence data. Firstly, iterative morphological processing was used to improve the contrast of low-resolution ship target profile edge features compared with the sea surface background. Next, the target area after adaptive segmentation was used to eliminate false alarms. As a result, the invalid background information of extensive comprehensive data was quickly eliminated. Then, support vector machine classification of S-HOG feature was carried out for suspected targets, and interference such as islands and reefs, broken clouds, and waves were eliminated according to the shape characteristics of ship targets. The method of multi-frame data association and searching for adjacent target information between frames was adopted to eliminate the interference of static targets and broken clouds with similar contours. Finally, the sequential marks were further trained and learned, and further false alarm elimination was completed based on the clipped Yolo network. Compared with the traditional Yolo Tiny V2/V3 series network, this method had higher computational speed and better detection performance. The F1 number of detection results was increased by 3%, and the calculation time was reduced by 66%.