60results about How to "Improve object detection performance" patented technology

The invention provides a multi-modal target detection method and device and a multi-modal recognition system. The method comprises the steps that a scene image is acquired, and the scene image is an image of a current environment acquired by adopting image acquisition equipment; acquiring a point cloud image, the point cloud image being an image of point cloud data of the current environment acquired by using a laser radar; according to the scene image and the point cloud image, fusion is carried out to obtain a target image, and the fusion mode at least comprises image fusion and / or feature fusion; and performing target identification on the target image to obtain a target object in the target image. According to the scheme, the problem of low accuracy of target detection based on single data in the prior art is solved.

The invention discloses a rainy day target detection method based on paired image data generation and knowledge distillation, and belongs to the technical field of automatic driving target detection. The method comprises the following steps: acquiring a real image of a road scene and a semantic segmentation mask; inputting the semantic segmentation mask and the weather condition text information into a trained improved diffusion model to obtain a road scene generation image under a rainy day condition; introducing a CoT module and an ODConv module into the target detection model for optimization, and then improving the target detection model by adopting a dynamic weighted aggregation loss function to obtain an optimized target detection model; performing knowledge distillation to obtain a rainy day target detection enhancement model; and performing target detection by using the rainy day target detection enhancement model. The method provided by the invention is excellent in performance in a rainy day scene, has good generalization ability, can provide higher-precision target detection in various complex scenes, and effectively improves the detection accuracy and stability.

The invention discloses a monitoring video target detection method based on multi-frame high-frequency difference enhancement, and the method comprises the following steps: 1, firstly, selecting a series of monitoring videos, carrying out the segmentation and frame extraction of the monitoring videos, and sampling one frame every ten frames as a training set, carrying out data preprocessing and enhancement operation on the input continuous image frames; and 2, constructing a multi-frame image reconstruction module to reconstruct an image according to the low-layer feature maps of the current frame and the reference frame, and optimizing parameters of the reconstruction module by calculating mean square error (MSE) loss between the original image and the reconstructed image. According to the monitoring video target detection method based on multi-frame high-frequency difference enhancement, a continuous frame high-frequency difference chart and a self-adaptive fusion enhancement feature strategy are ingeniously combined, high-frequency motion features are extracted and fused through reconstruction of continuous frames, and a powerful target detection framework is constructed; the target detection performance in the monitoring video can be obviously improved.

The invention provides a novel YOLOv10-based safety belt detection method and system, and the method comprises the steps: obtaining image data, and marking the image data, namely marking the position / wearing state of a safety belt; constructing a detection model based on a novel YOLOv10 network, namely replacing partial standard convolution Conv of the Backbone network with local attention regularization convolution; using a separable convolution star module to replace an original feature extraction module of the Backbone network; a CARAFE module is used for replacing up-sampling operation in a Neck network, multi-scale space attention dilation convolution fusion modules are additionally arranged before three sets of feature maps output by the Neck network, and then the feature maps are output to a Head network; and training a detection model by using the labeled image data to obtain a safety belt detection model. When the novel YOLOv10 network is used for safety belt detection, the detection efficiency and performance can be remarkably improved.

The invention discloses a 3D target detection method and system based on binocular vision and open vocabularies, and relates to the technical field of target detection. The method comprises the following steps: calculating pixel features of each image by adopting an open vocabulary 2D detection model; pixel features of the left view and the right view are aggregated to each three-dimensional point; obtaining a parallax estimation map based on the pixel features of the left and right views; introducing a time aggregation guide parallax consistency constraint to obtain a depth parallax estimation map; converting the depth parallax estimation map into a 3D point cloud, and extracting features to obtain 3D features; and carrying out correlation integration on the text features, the 2D fusion feature vectors and the 3D features by utilizing object-level image guide detection loss, object-level cross comparison alignment and object-level self-comparison learning to obtain a three-dimensional target detection model, realizing three-dimensional target detection and distinguishing different target objects. According to the invention, accurate detection and distinguishing of the target can be realized, so that the flexibility and universality of 3D target identification are improved.

The invention relates to the technical field of radar weak target detection, in particular to a radar weak target detection method and device based on deep learning. According to the technical scheme, the method comprises the following steps: data preprocessing and enhancement: receiving an original echo signal of a radar, generating three-dimensional time frequency-wavelet coefficient spatial data by combining time frequency analysis with wavelet transform, and expanding a data set in multiple modes; and feature extraction network construction: constructing a 3D-CNN and LSTM fused network, wherein the 3D-CNN has a convolution operation with a variable convolution kernel size. According to the method, the data are richer and more comprehensive through time-frequency analysis in combination with wavelet transform and data enhancement, and multi-dimensional features can be accurately captured through 3D-CNN and LSTM fusion and an adaptive convolution kernel; and the training effect is ensured through the adaptive learning rate and the early stop method, and the accuracy and adaptability of weak target detection are jointly improved.

The invention relates to the technical field of signal processing, and provides an airborne radar multipath clutter suppression method and device. The method comprises the following steps: calculating test statistics of each training sample in a sample set, and selecting a training sample corresponding to a median of the test statistics as a reference sample; calculating the mahalanobis distance between the reference sample and each training sample, carrying out the non-uniform measurement of each training sample through employing the mahalanobis distance, obtaining a uniform training sample through screening, and determining a to-be-detected unit sample where the multipath clutter is located; constructing an air-space domain clutter covariance matrix of the to-be-detected unit sample according to the to-be-detected unit sample, calculating the rank of the air-space domain clutter covariance matrix so as to construct a multi-stage Wiener filter and a blocking matrix, and constructing an air-space domain adaptive weight vector according to the multi-stage Wiener filter and the blocking matrix; and performing space-time joint processing according to the space-space domain adaptive weight vector to obtain an output signal. According to the invention, the target detection performance of the airborne early warning radar in the multipath clutter environment of the complex mountainous area is improved.

The invention belongs to the technical field of image processing, and particularly relates to a dual-light image target detection method under the view angle of an unmanned aerial vehicle, and the method comprises the steps: constructing a dual-light image target detection model which comprises a dual-light feature extraction and fusion network, a feature pyramid network and a rotating target detection head network; preprocessing the visible light and infrared light image data under the view angle of the unmanned aerial vehicle; inputting the preprocessed image pair into a dual-light feature extraction and fusion network to obtain fusion feature maps of different levels; inputting the fusion feature maps of different levels into a feature pyramid network, and outputting a multi-scale fusion feature map; the multi-scale fusion feature map is input into a rotating target detection head network, regression is carried out to obtain a rotating surrounding frame, a category and a confidence coefficient of the target under the view angle of the unmanned aerial vehicle, and the target detection capability in a complex environment is enhanced by a dual-light feature extraction and fusion network; the rotating bounding box regression strategy solves the problem that an existing horizontal bounding box is inaccurate in positioning when the target rotates.

The invention belongs to the technical field of radars, and discloses a radar clutter trace point screening method using target short-time motion characteristics. The invention provides a radar clutter trace point screening method based on target motion characteristic parameter constraint and a short-time stability principle from trace point layer research on the basis that target motion generally has continuous and stable space-time relevance and clutter trace points appear randomly. According to the method, a set of clutter trace point screening method rules is provided based on target motion characteristic parameter constraints and short-term stability of target motion by only utilizing time and position basic information carried by trace points, clutter trace points can be effectively eliminated in a dense clutter background, target trace points can be effectively extracted, and the target trace points can be effectively extracted. And an effective technical support is provided for radar weak target detection and tracking under the clutter background.

The invention provides a communication perception integrated waveform design method for the Internet of Vehicles, which belongs to the technical field of wireless communication, and can realize peak performance, enhance radar capability and ensure efficient communication by minimizing the weight sum of multi-user interference, the similarity between a design waveform and a reference waveform, a Cramer-Rao bound and an ISAC waveform. According to the method, optimization with the total transmission power and the peak-to-average power ratio as constraints is adopted, a non-convex problem is converted into a convex problem, an alternating direction multiplier method algorithm is adopted for solving, and a simulation result shows that the ISAC waveform has good communication and target detection capability, and the V2X spectrum utilization rate can be improved. According to the method, a target can be detected during communication with a downlink user, the sum of MUI, CRB and waveform similarity is minimized, weighing parameters are introduced, flexible weighing between communication and sensing is achieved under the constraint of the total power and the peak-to-average power ratio, the method better adapts to a V2X high-dynamic scene, and mutual interference caused by separation of traditional sensing and communication functions is reduced.

The invention relates to a turnout detection method based on computer vision and an intelligent monitoring pile. In order to improve the detection precision, an improved YOLOv8 model is provided, an original path aggregation feature pyramid network structure is jointly improved by using a space-depth convolution module (SDC) and a feature enhancement-fusion module (FEF) in a check layer of the network, and a small target detection enhancement pyramid structure is provided to enhance the small target detection capability. In addition, an improved loss function with a scaling factor and an auxiliary bounding box is used to further enhance the detectability of the model. Compared with the prior art, the method has higher accuracy and robustness.

The application provides a point cloud augmentation method and device, a computer readable storage medium and a terminal device. The point cloud augmentation method comprises: obtaining original point cloud data and at least one image, the original point cloud data and the at least one image being collected for the same target; projecting each original point cloud in the original point cloud data into the at least one image to obtain a plurality of projection points in the at least one image, each projection point corresponding to an original point cloud; selecting at least one adjacent pixel as an extended pixel in a vertical direction of each projection point according to a preset interval; and at least mapping each extended pixel to a three-dimensional space to obtain an augmented point cloud. The application can improve the number of point clouds and the effectiveness of the augmented point cloud.

The present application relates to the radar signal processing neighborhood, specifically to a kind of target detection method based on target space-time stability.The method includes: based on the time-domain signal representation of multiple frames sliding, generate subframe by dynamic matching filter, obtain subframe peak point information, including the energy, distance, doppler information of peak point on each subframe RD spectrum;Based on the spatial signal processing of single snap angle, generate multi-subframe peak point RDA spectrum, obtain the azimuth information of peak point on each subframe RD spectrum, utilize the feature information of each subframe peak point, remove false alarm point using false alarm suppression method based on mahalanobis distance;After false alarm suppression, the result is further removed discrete false alarm point using target track based on space-time stationarity, the method proposed in the present application improves the problem of too many false alarms in the RD spectrum after traditional clutter suppression algorithm;It can effectively reduce false alarm rate while retaining weak target signal, improve target detection capability;It is suitable for multiple system radars, and the application range is wide.

The embodiment of the application relates to an algorithm and a device of a Bayesian classifier of a non-uniform background, which are applied to an underwater active sonar system, the algorithm comprising: obtaining underwater data to be measured and auxiliary data through the active sonar system; the number K of the auxiliary data is greater than 0; modeling classification of the unknown covariance matrix structure into a binary hypothesis testing problem; hypotheses of the binary hypothesis testing problem comprise H i Wherein i=0, 1, H0 is a case that the unknown covariance matrix is a complex conjugate symmetric matrix; H1 is a case that the unknown covariance matrix is a real symmetric matrix; a Bayesian model is set, the Bayesian model comprising a complex inverse Wishart random matrix and a real inverse Wishart random matrix; a classifier for distinguishing the two hypotheses is obtained by using a minimum error probability criterion.

This invention discloses a radar target detection method based on graph node dual-channel feature attention fusion, belonging to the field of radar signal detection technology. The method includes the following steps: Step 1: Divide the received frame radar echo data into graph nodes; Step 2: Extract time-domain amplitude and time-frequency features from the echo time-series data corresponding to each graph node; Step 3: Construct a feature preprocessing subnetwork; Step 4: Construct a node feature fusion subnetwork; Step 5: Construct a signal classification graph neural network; Step 6: Connect the feature preprocessing subnetwork, the node feature fusion subnetwork, and the signal classification graph neural network in series to form a radar target detection neural network; Step 7: Input the test set into the trained radar target detection neural network and output a binary classification result indicating whether the corresponding node is a target or clutter signal. This method can improve the radar's target detection capability in clutter environments.

The present application relates to the field of active sonar target detection, and relates to a kind of adaptive target detection method and system suitable for reverberation edge environment.The method of the present application comprises: selecting the data of unit to be detected as main data, and selecting the received data of multiple distance units on both sides of the unit to be detected as overall auxiliary data;Overall auxiliary data is distributed in the three uniform reverberation areas formed by the left reverberation edge and the right reverberation edge of the unit to be detected, and is divided into three auxiliary data subsets accordingly;Modeling target adaptive detection as a binary hypothesis testing problem including no-target hypothesis and target hypothesis;Obtain the joint probability density function of main data and overall auxiliary data under no-target hypothesis and target hypothesis;Based on the modified two-step generalized likelihood ratio test criterion, the joint probability density function and the probability density function of main data and auxiliary data subsets are processed in turn, and the final detection expression is obtained to realize target adaptive detection.

The invention relates to an inductance-pass integrated hybrid beam forming method based on a one-bit digital-to-analog converter. The method comprises the following steps: designing an analog precoder and a one-bit digital-to-analog converter quantized signal passing through a digital precoder based on a trade-off optimization problem between radar and communication performance; a transmitting beam vector is designed according to the designed analog precoder and the quantized signal of the one-bit digital-to-analog converter, an optimization expression of the transmitting beam vector is solved, and an optimal transmitting beam vector is obtained; and designing an optimization problem of quantized signals of the analog precoder and the one-bit digital-to-analog converter according to the optimal transmitting beam vector, solving the optimization problem by using a Matlab toolbox and a BTSA-GA algorithm, and generating the sensing and communication integrated hybrid beam according to the optimized quantized signals of the analog precoder and the one-bit digital-to-analog converter. By adopting the method, the complexity and the power consumption of the large-scale MIMO system can be reduced.

The invention discloses a crack detection method and system based on edge end equipment, and the method comprises the following steps: collecting an image frame of a crack condition to be detected, and carrying out the preprocessing of the image frame; performing matrix construction on the preprocessed data, and performing image feature extraction on the preprocessed data by using convolution calculation; performing quantization processing on a feature extraction result by using an INT8 quantization algorithm; inputting the quantized data into a Leaky ReLU function to carry out data nonlinear transformation; constructing a 2X2 data matrix, and processing the data after nonlinear transformation by using a maximum pooling method; carrying out fusion operation on the feature maps of different levels; performing further feature extraction on the fused feature data to obtain output layer data; analyzing the output layer data; and converting the target frame information into RGB data for visual display. The target detection effect can be improved, and the problems of large model size, large equipment weight, high system power consumption, dependence on a network and the like in the existing field environment crack detection technology are solved.

The invention relates to the field of radar signal processing, in particular to a method for estimating the elevation angle of a low-altitude target. According to the method, different terrain conditions are comprehensively considered, the RML algorithm and the multi-frequency beam scanning algorithm are fused for processing, in the flat terrain environment, the RML algorithm is adopted to be combined with signal space-time information, distance information and a prior reflection coefficient, a composite steering vector under the multi-path condition is generated to estimate the elevation angle of the target, and the elevation angle of the target is estimated. The estimation mode utilizes various prior information and has the characteristics of low calculation amount and high angle measurement precision, and aiming at the fluctuating terrain environment, a multi-frequency beam scanning algorithm is adopted, so that the angle measurement precision interference caused by relatively strong ground diffuse reflection components can be effectively avoided.

The invention discloses a severe environment unmanned aerial vehicle aerial photography target detection method based on multispectral iteration enhancement, and belongs to the technical field of computer vision. The method comprises the following steps: firstly, acquiring RGB pictures and infrared pictures in severe environments such as rain and fog, night and the like, and constructing a data set; then, an unmanned aerial vehicle aerial photography target detection model is constructed, and efficient deep fusion of visible light and infrared multispectral features is realized by introducing a novel iterative hierarchical attention and differential enhancement fusion framework; meanwhile, a tail enhancement unit of an HGBlock module in the backbone network is redesigned; then training an unmanned aerial vehicle aerial photography target detection model by using the data set; and finally, using the trained detection model to detect an aerial target of the unmanned aerial vehicle and evaluate the performance of the model. Through a fusion mode of combining interactive alignment, differential enhancement and iterative feedback, the target detection performance and real-time performance of the unmanned aerial vehicle under complex weather and illumination conditions are improved.

This application discloses a few-shot target detection and recognition method based on self-supervised subspace separation, belonging to the field of remote sensing image processing technology. The proposed innovative framework completely isolates the training of base classes and novel classes at three levels. First, a low-rank subspace adapter is proposed for structural separation, achieving network optimization for novel classes without compromising the pre-training performance of base classes. This uses fewer parameters to modulate network components, alleviating the overfitting problem caused by fully fine-tuning under few-shot conditions. Second, an orthogonal subspace extractor is proposed for feature decoupling, adaptively learning the corresponding subspace for each class and extracting decoupled representations, enhancing the separability between classes. Third, a balanced classifier is designed for loss balancing, preventing the final prediction result from being overly biased towards the background or base class. The framework proposed in this invention has significant advantages in few-shot target detection and recognition, showing significant improvements in both base and novel classes.

The invention belongs to the technical field of image processing, and relates to a personnel falling-into-water detection method and system and a readable medium, and the method comprises the following steps: obtaining a video or image of a to-be-detected water surface; segmenting the video or image of the water surface through an improved YOLOv8 model to obtain a water surface mask; performing pedestrian detection in the video or image of the water surface through a YOLOv7 model; and if the YOLOv7 model detects a pedestrian and the pedestrian is in the segmented water surface mask, judging that a person is drowning, otherwise, judging that no person is drowning. According to the scheme, the multi-scale features can be effectively represented and processed, so that the efficiency and accuracy of personnel falling-into-water detection are improved.

The present invention discloses a method and electronic device for detecting a drone target, wherein the method comprises: obtaining an image sequence to be detected according to video data to be detected; extracting dynamic region images and static space images corresponding to image frames in the image sequence; fusing dynamic region features and static space features, and performing multi-template matching to determine whether a drone target exists in the image frame. The present invention can improve the detection accuracy of low-speed, slow, and small drones.

The invention provides a wide-angle optical lens which can solve the problems that an existing wide-angle optical lens is small in target surface and difficult to have a large view field and low distortion at the same time. The lens comprises a first meniscus positive lens, a first meniscus negative lens, a second meniscus negative lens, a plano-concave lens, a third meniscus negative lens, a first biconvex positive lens, a second meniscus positive lens, a biconcave negative lens and a second biconvex positive lens which are sequentially arranged from an object plane to an image plane along the propagation direction of incident light, the focal length of the first positive meniscus lens is f1, the focal length of the first negative meniscus lens is f2, the focal length of the second negative meniscus lens is f3, the focal length of the plano-concave lens is f4, the focal length of the third negative meniscus lens is f5, the focal length of the first biconvex positive lens is f6, the focal length of the second positive meniscus lens is f7, and the focal length of the biconcave negative lens is f8. The focal length f9 of the second biconvex positive lens and the focal length f of the integral structure formed by the nine lenses meet the following relations: 2.60 lt; f1 / flt; 3.60,-1.20 [mu] m; f2 / flt; 1.00 [mu] m, 1.00 [mu] m; f3 / flt; -0.80,-1.80 [mu] m,-0.80 [mu] F4 / flt; 2.50,-0.4 lt, 2.50,-0.4 lt; f5 / flt; 0.2, 0.2 lt, 0.2 lt; f6 / flt; 0, 0.4, 0.70 lt; f7 / flt; 0.90,-0.90 < lt >,-0.90 < lt >; f8 / flt; 0.60, 1.00 lt, 0.60, 1.00 lt; f9 / flt; 1.20, 1.20.

The application discloses a target detection method based on motion and event information decoupling, aiming at the problem of coupling of appearance information and motion information in an event stream, proposes an event representation method for decoupling motion and appearance, and designs a double-flow detection network and an event motion guided attention module based on decoupled input, uses motion features to enhance appearance features, and thus greatly improves target detection performance.

The application provides an open-vocabulary aerial image target detection method based on collaborative quality perception, constructs a labeled basic category dataset and an unlabeled new category dataset of aerial images, constructs an SQAPN target detection model, the target detection model comprises a backbone network and an RPN network with a spatial alignment perception module, an alignment consistency loss function is introduced in the training process of the target detection model to supervise the training of the spatial alignment perception module, a confidence modulation module is introduced in the inference stage to obtain a comprehensive quality score, the target detection model is trained by using the dataset to obtain a trained target detection model, in the inference stage, the aerial image to be processed is input into the trained target detection model, and a comprehensive quality score and a candidate box are output, candidate box screening is completed through non-maximum suppression (NMS), and finally, the detection result is output. The application can effectively improve the open-vocabulary aerial image target detection precision and robustness.

The invention discloses an SAR image ship target detection method, device and equipment and a medium, and belongs to the technical field of artificial intelligence. The method comprises the steps of segmenting an SAR image into a plurality of sub-images based on a slice-assisted super reasoning algorithm; the multiple sub-images are input into a target detection model, a ship target detection result in each sub-image is acquired, and the target detection model is a pre-trained convolutional neural network model; and combining the ship target detection results of the plurality of sub-images to obtain a ship target detection result in the SAR image. According to the ship target detection method provided by the invention, a C2f module of a dynamic snakelike convolution improved YOLOv8n backbone network is introduced, different depth features output by the backbone network are fused by using an asymptotic feature pyramid network, and ShapeIOU is adopted as a loss function, so that the detection capability of a target detection model on targets with different scales, shapes and postures is enhanced, and the detection accuracy of the target detection model is improved. Therefore, the accuracy and robustness of small and medium-sized ship target detection of the SAR image are improved.

This invention discloses a lightweight star catalog target detection method and system. The method includes: constructing a student model, which employs a lightweight feature extractor and a Transformer decoder with width and depth pruning; constructing a teacher model, which employs an un-lightweight feature extractor and a Transformer decoder without width and depth pruning; adjusting the parameters of the teacher and student models, calculating the value of the loss function, stopping the iteration when the value of the loss function is minimized, and obtaining trained teacher and student models; and using the trained student model to perform lightweight star catalog target detection. The advantages of this invention are: lightweight, low computational complexity, and the ability to achieve 3D position detection of star catalog targets.

The application discloses a flood discharge detection model training method and device, equipment and a medium, the method comprises the following steps: taking the video image data of the pre-collected reservoir spillway as a first training sample set, and performing sample enhancement on the first training sample set according to a preset sample enhancement strategy to obtain a second training sample set; wherein the sample enhancement strategy comprises one or more combinations of the following: an independent enhancement strategy, a combined enhancement strategy or an inlay enhancement strategy; training a pre-constructed initial flood discharge detection model according to the second training sample set to obtain a final flood discharge detection model; wherein the initial flood discharge detection model is an improved YOLOv8 model obtained by modifying a flood discharge convolution feature extraction module and updating a feature fusion module. The application can improve the calculation efficiency and target detection accuracy of the flood discharge detection model in a small sample data scene.

The present invention relates to the technical field of image illumination enhancement, and specifically discloses a method for target detection based on adaptive feature learning of low-light images, the steps of which include inputting a target image into YOLOv5 and Zero_DCE respectively, obtaining features after removing illumination influence and target features, and fusing the two to obtain a result image; the fusion is as follows: the image processed by Zero_DCE is input into a backbone network for convolution to obtain a first intermediate result, and YOLOv5 convolves the image and upsamples it to obtain a second intermediate result, the first intermediate result and the second intermediate result have the same image size, and the two are fused. The present invention proposes a parallel target detection algorithm under low-light conditions, which uses an illumination enhancement network and a target detection network to encode image features in parallel, and the operation includes fusing illumination-related features with features for specific target detection to provide sufficient semantic information for subsequent detection tasks.