259 results about "Large target" patented technology

The invention discloses a low-scattering support system design and optimization method for large-scale target external field measurement, the support system comprises a support and a low-scattering end cap, and the low-scattering end cap is connected with the top of the support; the method comprises the following steps: S1, constructing a bracket geometric model based on a low-scattering shape design formula, and evaluating the bearing capacity of a bracket; s2, performing electromagnetic solution on the bracket by using a non-conformal region decomposition area integral equation method, and optimizing a bracket geometric model; s3, constructing a low-scattering end cap based on a low-scattering shape design formula; and S4, placing the low-scattering end cap at the top of the bracket to form a low-scattering support system, and evaluating the composite backscattering of the low-scattering support system by using a non-conformal region decomposition area integral equation. According to the low-scattering supporting system, high bearing capacity and good low-scattering performance are both considered, and reliable supporting guarantee can be provided for external field measurement of a large target.

The utility model provides an all-glass vehicle-mounted day and night confocal optical system and a camera module applying the same, mainly comprising eight lenses, the first lens has negative focal power, the object side surface is a convex surface, and the image side surface is a concave surface, the second lens has negative focal power, the object side surface is a concave surface, and the image side surface is a convex surface. The third lens has positive focal power, the object side surface is a convex surface, the image side surface is a concave surface, the fourth lens has negative focal power, the object side surface is a convex surface, the image side surface is a concave surface, the fifth lens has positive focal power, the object side surface is a convex surface, the image side surface is a convex surface, and the fourth lens and the fifth lens form a bonding lens. The sixth lens has positive focal power, the object side face of the sixth lens is a convex face, the image side face of the sixth lens is a convex face, the seventh lens has negative focal power, the image side face of the seventh lens is a concave face, the eighth lens has focal power, and the image side face of the eighth lens is a concave face. And the picture illumination uniformity and the imaging definition are improved.

The invention discloses an SAR directed target detection method based on multi-scale context sensing, and belongs to the technical field of remote sensing image processing and computer vision. According to the method, firstly, through a scattering characteristic guided multi-scale dynamic characteristic enhancement network, a multi-branch structure and a dynamic fusion mechanism are utilized to enhance characteristic expressions of targets of different scales; and secondly, designing a task-adaptive regional context sensing module, fusing local details and semantic contexts, and improving the target discrimination capability under a complex background. And a decoupling type progressive refining detection head is adopted to predict a target category, a directed bounding box and a direction angle, and the positioning precision is optimized through progressive regression. Semantic consistency loss is introduced in the training stage, and end-to-end optimization is achieved in combination with a multi-task joint strategy. According to the method, the problems of large target scale difference, strong background interference, changeable directions and the like in the SAR image are effectively solved, and the accuracy and robustness of directed target detection are remarkably improved.

The invention relates to a light and miniature big-target-surface low-light lens, and the lens comprises a front lens group A, a diaphragm B and a rear lens group C, wherein the front lens group A, the diaphragm B and the rear lens group C are sequentially arranged in an incident direction of an optical path. The front lens group A includes a positive meniscus lens A1, a negative meniscus lens A2,a biconcave lens A3 and a biconvex lens A4, wherein the positive meniscus lens A1, the negative meniscus lens A2, the biconcave lens A3 and the biconvex lens A4 are sequentially arranged in an incident direction of the optical path. The biconcave lens A3 and the biconvex lens A4 are in close connection with each other as a first bonding group. The rear lens group C includes a biconcave lens C1, abiconvex lens C2, a biconvex lens C3, a plano-convex lens C4 and a negative meniscus lens C5, wherein the biconcave lens C1, the biconvex lens C2, the biconvex lens C3, the plano-convex lens C4 and the negative meniscus lens C5 are sequentially arranged in an incident direction of the optical path. The biconcave lens C1 and the biconvex lens C2 are in close connection with each other as a secondbonding group. The invention also relates to an imaging method of the lens. The lens is reasonable in design, effectively improves the imaging quality under the requirement of the large target surfaceof the lens, reduces the number of optical lenses, and realizes light and miniature lens. The lens has strong light collecting and light transmitting ability, and the target surface of the lens is large. The lens is high in optical resolution, is large in angle of view, and is large in observable range.

The invention relates to a 1.2-inch intelligent traffic monitoring system lens and a working method thereof. An optical imaging system of the lens is sequentially provided with a first lens to a fourteenth lens from an object space to an image space. Wherein the focal power of each lens adopts a positive and negative combined form, so that optical aberrations such as spherical aberration and the like are greatly reduced. Wherein the fourth lens and the fifth lens form an achromatic doublet lens group, the seventh lens and the eighth lens form an achromatic doublet lens group, the ninth lens and the tenth lens form an achromatic doublet lens group, and the eleventh lens, the twelfth lens and the thirteenth lens form an achromatic triplet lens group. Through reasonable lens matching, an optical system realizes large aperture, large target surface and no defocus in day and night, and through good correction of on-axis and off-axis aberrations, the lens has good imaging quality.

The invention belongs to the technical field of information remote sensing, and particularly discloses a port and bridge large-scale target thematic map generation method based on an SAR technology, and the method comprises the steps: obtaining multi-temporal synthetic aperture radar image data of a port region, processing the original signal intensity distribution of the image data through denoising and filtering, and obtaining a smoothed scattering intensity map; according to the scattering intensity graph, extracting echo amplitude and phase information of a target area, fusing multi-polarization channel data, and determining a scattering characteristic parameter set; grouping the scattering characteristic parameter set based on a clustering algorithm, and separating preliminary clustering results of facility classes and residue classes to obtain feature vector representation; matching the feature vector representation with a stored shape template according to a pre-established feature database; the objective of the invention is to solve the problem in the prior art that environment management is difficult to accurately position a business scene due to mixed distribution of facilities and residues in a port area.

The invention relates to a refraction-super hybrid long-focus day and night confocal lens and an image acquisition device, and belongs to the field of optical lenses. The long-focus day and night confocal lens has focus consistency in a visible light band and an infrared band; comprising a first lens, a second lens, a third lens, a fourth lens, a fifth lens, a sixth lens, a seventh lens and an eighth lens which are sequentially arranged from an object plane to an image plane along an optical axis; at least one of the first lens, the second lens, the third lens, the fourth lens, the fifth lens, the sixth lens, the seventh lens and the eighth lens is a metasurface lens, and the other lenses are spherical lenses. By adopting the technical scheme of folding and super mixing, large-target-surface day and night common-focal-length imaging is realized, the use of glued lenses is reduced, the complexity of the system is reduced, the assembly difficulty is reduced, and the risk that a structural member generates stray light is reduced.

The invention relates to a small-sized light-weight large-target-surface infrared athermalization optical system, and the optical system consists of a first meniscus positive lens, a meniscus negative lens, a second meniscus positive lens and a third meniscus positive lens which are sequentially and coaxially arranged from an object side to an image side, the second meniscus positive lens and the third meniscus positive lens are the same lens, the first meniscus positive lens, the second meniscus positive lens and the third meniscus positive lens are all made of chalcogenide glass IRG206, and the optical system ensures that edge imaging is clear and the image surface diameter reaches 20mm through optimal arrangement of focal power and optical materials of all the lenses and intervals among all the lenses, so that the image quality is improved. The long-wave infrared imaging device can adapt to a 1280 * 1024 array and a long-wave uncooled infrared detector with the pixel size of 12 microns, and has the advantages of high imaging quality, small size and light weight.

The invention relates to the technical field of remote sensing image processing and computer vision, in particular to a mining area earth surface feature remote sensing semantic segmentation method and device, electronic equipment and a storage medium. The high-resolution remote sensing image is a high-resolution No.2 original image containing blue light, green light, red light, near-infrared and panchromatic wave bands. A remote sensing semantic segmentation model of the surface features of the mining area is constructed, and a ResNeSt module and an SE channel attention module are introduced into an encoder, so that the feature response and discrimination capability of the model to slender and curved boundaries of key elements such as a dam body and a reservoir area of the mining area is effectively enhanced. And meanwhile, multi-scale context information is aggregated in combination with an ASPP module, so that the problems of boundary fracture and detail loss caused by large target scale difference are effectively solved, and the extracted mining area edge contour is more continuous and complete and is remarkably superior to a traditional segmentation method.

The present application provides an optical lens (110), a camera module (100), and an electronic device (10). The optical lens (110) comprises a catadioptric lens (111), a lens group (112) and a light folding member (113). An incident light ray enters a first transmission surface (111a) of the catadioptric lens (111) and is refracted therethrough, then is reflected by a first reflection surface (111b) and a second reflection surface (111c) in sequence, and is refracted through a second transmission surface (111d) and is emitted; after passing through the lens group (112), the light ray is reflected multiple times in the light folding member (113) and then is emitted. The catadioptric lens (111) can achieve two-time light reflection and large aperture, the lens group (112) can achieve light ray concentration and adjustment, and the light folding member (113) can achieve multi-time light reflection. The optical lens (110) has a long back focal magnification light path and a relatively large equivalent focal length, and can be used in combination with an image sensor (120) having a large target surface, thereby achieving high-precision and high-resolution shooting. A folded light path utilizing the catadioptric lens (111) and the light folding member (113) can achieve small height and volume of the optical lens (110) and a relatively small shoulder height. The camera module (100) comprises the image sensor (120), a focusing motor (130), an anti-shake motor (140) and the optical lens (110). The electronic device (10) comprises a device housing (200) and the camera module (100).

The application discloses a multi-target pose real-time online estimation method, system, device and medium, and aims at real-time aerial images. On one hand, the ground target pose is quickly estimated through block detection and head-to-tail interconnection. On the other hand, the ground target is tracked in real time online in combination with a data association thought. The problems of small target volume, large target quantity, easy to be shielded, dynamic non-cooperation and limited calculation capacity of embedded equipment in the aerial images are effectively overcome. The efficiency of aerial multi-target pose estimation is improved. The aerial ground target online real-time estimation in the airborne embedded environment is realized. The multi-target, multi-dimensional and strong real-time requirements of ground target detection and sensing in the fields of disaster rescue and military monitoring are met.

The utility model discloses a scanning device for three-dimensional model information acquisition, and relates to the technical field of three-dimensional scanning, and the scanning device comprises a mainframe box which is internally provided with a lifting driving unit; the lower sliding column is inserted into the mainframe box, and a lower sliding groove is formed in the side face; the upper sliding column is rotationally connected to the upper end of the lower sliding column, an upper sliding groove is formed in the side face of the upper sliding column, and the upper sliding groove communicates with the lower sliding groove; a lower movable groove is formed in the lower sliding column; an upper movable groove is formed in the upper sliding column, the lower movable groove and the upper movable groove form an annular sliding groove, and an annular chain is wound in the annular sliding groove; the lifting driving unit drives the chain to move in the annular sliding groove. The sliding rod is inserted into the lower sliding groove and connected with the chain, and a scanning unit is arranged at the tail end. The sliding column on the mainframe box is divided into the lower sliding column and the upper sliding column, and the upper sliding column can rotate by 90 degrees, so that the scanning unit can scan a large target and a small target in different postures, and the scanning requirements of various different targets are met.

The present application relates to the technical field of unmanned aerial vehicle, and discloses a field splicing method of distributed photoelectric tracking system, first, a frequency spectrum monitor is used to find an unmanned aerial vehicle target and send the azimuth and distance data of the target to a target detection unit, after the target is found by the target detection unit, the target is automatically or manually locked, and the information of the locked target is transmitted to a target tracking unit, the target tracking unit extracts features in the area around the original position of the target in the last frame, finds the position with the highest matching degree of the target features and takes the position as the new position of the target, then adjusts the camera to the new position, automatically or manually controls the linked interference device, and transmits satellite navigation and remote control interference signals to the target area until the unmanned aerial vehicle is driven away. The present application solves the problem of too large target position error of the frequency spectrum monitor, and has accurate positioning, and meets the real-time and success rate requirements of target search.

The utility model relates to a large-target-surface long-focal-length two-variable high-definition zoom lens, which comprises a lens, and an optical system of the lens comprises a front fixed lens group, a zoom lens group, a compensating lens group and a rear fixed lens group which are sequentially arranged from left to right along a light incidence direction. According to the utility model, the design is reasonable, a continuous zoom closed-loop control function, a focusing closed-loop control function, a focal length feedback function and an image output function are realized, large-range searching and tracking camera shooting can be carried out on targets with far and near distances and high-speed movement, and the system is applied to scenes with requirements such as unmanned monitoring, early warning, marginal and coast defense and the like.

The application provides a target detection method based on group Hungarian matching and loU perception, and relates to the technical field of image detection. The target detection method based on group Hungarian matching and loU perception adopts multi-scale Hungarian matching and multi-scale query selection, avoids the situation that final prediction targets are all inclined to large targets, and improves small target precision; meanwhile, an IoU branch is added in the training stage, and IoU confidence and classification confidence are combined as a comprehensive confidence evaluation index, in the Topk stage, higher-quality initial anchor frames can be selected, the model can be fitted faster, and therefore higher precision improvement can be achieved at 12 epochs.

The application discloses a magnetron sputtering coating method and a magnetron sputtering coating device, and belongs to the technical field of glass substrate sputtering. The method comprises the following steps: obtaining a first area of a target material sputtering surface; constructing an auxiliary magnetic field vertically towards the surface of a substrate in a coating chamber when a magnetron moves to a position vertically opposite to other areas of the sputtering surface except the first area; and closing the auxiliary magnetic field when the magnetron moves to a position vertically opposite to the first area. The speed of sputtering particles flying to the substrate in the edge area can be effectively improved, thereby improving the efficiency. When the magnetron is vertically opposite to the first area, the auxiliary magnetic field is closed, the through-hole sidewall coverage and film thickness uniformity of the edge area of the substrate are significantly improved on the premise of maintaining a high deposition rate in the central area, and the inherent defect of 'thin edge and thick center' in a large target-substrate distance process is solved.

The application discloses a high-resolution oblique-axis optical lens, comprising, from the object side to the image side, a first lens, a second lens, a third lens, a fourth lens, a diaphragm, a fifth lens, a sixth lens, a seventh lens, an eighth lens, a ninth lens, a protective glass and an image sensor; the first lens is a filter; the second lens is a biconvex spherical lens with positive focal power; the third lens is a spherical lens with negative focal power; the fourth lens is a spherical lens with positive focal power; and the fifth lens is a biconcave spherical lens with negative focal power. The technical scheme balances and corrects optical aberrations of all levels by reasonably distributing the focal power of the lenses, reasonably matching the optical materials and shapes, and utilizes the cylindrical lens to correct astigmatism caused by the optical path of the meridian arc vector plane of the protective glass, so that the advantages of large target surface, low distortion, ultra-short object distance, high resolving power and the like of the lens are achieved.

The application provides an optical lens, which comprises six lenses in sequence along an optical axis from an object side to an imaging surface, and the six lenses comprise: a first lens with positive focal power, wherein the object side surface of the first lens is a concave surface, and the image side surface of the first lens is a convex surface; a second lens with negative focal power, wherein the object side surface of the second lens is a concave surface, and the image side surface of the second lens is a convex surface; a third lens with positive focal power, wherein the image side surface of the third lens is a convex surface; a fourth lens with positive focal power, wherein the object side surface of the fourth lens is a convex surface; a fifth lens with negative focal power, wherein the image side surface of the fifth lens is a concave surface; and a sixth lens with positive focal power. The optical lens provided by the application can improve the imaging quality of the optical lens, reduce aberration, improve the imaging quality of the optical lens, and make the lens have one or more advantages such as a large target surface, a large aperture, high imaging quality and the like.

The application discloses a middle-wave refrigeration infrared lens, which comprises a first lens, a second lens, a third lens and a fourth lens arranged in sequence along the transmission direction of the optical axis; the first lens and the third lens are both meniscus positive lenses with the convex surface facing the object side; the second lens is a meniscus negative lens with the convex surface facing the object side; the fourth lens is a meniscus negative lens with the convex surface facing the image side; the air gap between the first lens and the second lens is 1 mm; the air gap between the second lens and the third lens is 54 mm; and the air gap between the third lens and the fourth lens is 5 mm; wherein the lens further comprises a reflecting mirror arranged between the second lens and the third lens to change the direction of the optical axis; the lens provided by the application has the advantages of small lens length size, simple structure, large target surface, clear imaging, and the ability to match a 320*256 resolution and a 30 mu m refrigeration type detector; and the application further discloses an imaging device.

The invention provides an optical lens. The optical lens comprises a first lens, a second lens, a third lens, a fourth lens, a fifth lens, a sixth lens and a seventh lens which are sequentially arranged from an object space to an image space along an optical axis, the first lens has negative focal power, the object side surface of the first lens is a convex surface, and the image side surface of the first lens is a concave surface; the second lens has negative focal power, the object side surface of the second lens is a concave surface, and the image side surface of the second lens is a convex surface; the third lens has positive focal power, and the object side surface of the third lens is a convex surface; the object side surface of the fourth lens is a convex surface, and the image side surface of the fourth lens is a concave surface; the fifth lens has positive focal power, and the object side surface and the image side surface of the fifth lens are convex surfaces; the sixth lens has negative focal power, and the object side surface and the image side surface of the sixth lens are concave surfaces; the seventh lens has negative focal power, the object side surface of the seventh lens is a concave surface, and the image side surface of the seventh lens is a convex surface. The invention has the advantages of small volume, low cost, large target surface, high pixel and high transmittance.

The application provides an optical lens, which comprises seven lenses in sequence along an optical axis from an object side to an imaging surface, and the seven lenses comprise: a first lens with negative focal power, wherein the object side surface is a concave surface, and the image side surface is a concave surface; a second lens with positive focal power, wherein the object side surface is a convex surface, and the image side surface is a convex surface; a third lens with negative focal power, wherein the object side surface is a concave surface, and the image side surface is a convex surface; a fourth lens with positive focal power, wherein the object side surface is a concave surface, and the image side surface is a convex surface; a fifth lens with positive focal power, wherein the object side surface is a convex surface, and the image side surface is a convex surface; a sixth lens with negative focal power, wherein the object side surface is a concave surface, and the image side surface is a concave surface; and a seventh lens with positive focal power, wherein the object side surface is a convex surface, and the image side surface is a concave surface. The optical lens provided by the application can improve the imaging quality of the optical lens, reduce aberration, improve the imaging quality of the optical lens, and make the lens have one or more advantages such as a large target surface, a large aperture, high imaging quality and the like.

The invention provides an optical lens, which comprises six lenses from an object side to an imaging surface along an optical axis: a first lens with positive focal power, a second lens with negative focal power, a third lens with positive focal power, a fourth lens with negative focal power and a fifth lens with positive focal power, the object side surface of the second lens is a concave surface, and the image side surface of the second lens is a convex surface; the object side surface of the third lens is a convex surface; the fourth lens has negative focal power; the fifth lens has focal power; the sixth lens has negative focal power; wherein the curvature radius R4 of the image side surface of the second lens and the curvature radius R5 of the object side surface of the third lens meet the formula: (R4 + R5) / (R4-R5) lt; 1. According to the optical lens provided by the invention, the six lenses with specific focal power are adopted, and through specific surface shape matching and reasonable focal power distribution, the imaging quality of the optical lens can be improved, the aberration can be reduced, the imaging quality of the optical lens can be improved, and the lens has one or more advantages of long focus, large target surface, high imaging quality and the like.

The application provides a target scattering characteristic test method based on support structure pointing difference servo, and belongs to the technical field of low observability. The method is to reduce the coupling relationship between the support rod and the target by making the low scattering support rod produce a known and certain difference rotation when rotating, so as to reduce the influence of the coupling relationship between the support rod and the target on the test precision. The application can reduce the influence of the coupling relationship between the support rod and the target on the test precision on the basis that the overall electromagnetic characteristics of the test background are in a lower order, and quickly complete the scattering characteristic test of the low scattering target. The support rod produces a known and certain difference servo motion track when the rotary table rotates, the coupling characteristics between the support rod and the target are reduced through track fitting, higher target scattering characteristic measurement precision is obtained, the requirement for the low scattering characteristics of the support rod is greatly relieved, and the scattering characteristic measurement accuracy of the large target is improved. The application is suitable for all test sites using the support rod to arrange the target in the electromagnetic wave irradiation area.

The present case relates to a kind of street scene infrared target detection method, after training, the model is too complex and the effect of extracting target feature from complex background such as street scene is not good for existing target detection model, the feature information of specified target is obtained by inputting the real-time acquisition street scene infrared image into infrared target detector, and the target is positioned and identified.The infrared target detector adopts YOLOv5 network and parameter-free attention module SimAM to be combined, realizes each repeated part of C3 residual module in YOLOv5 network structure, parameter-free attention module SimAM is added in series behind each repeated part, while not increasing parameter quantity, the purpose of suppressing background noise and enhancing detection performance is realized.Furthermore, different dilated convolution module of expansion rate is added after each SPPF module maximum pooling operation, so that the model obtains flexible receptive field, and the performance of detecting the target with large size difference is enhanced.

The invention discloses a large-view-field low-distortion infrared telecentric scanning objective lens which comprises a lens body, the lens body is composed of a first lens, a second lens, a third lens, a fourth lens and a fifth lens which are arranged along an axis, the first lens is a positive lens, the second lens is a negative lens, the third lens is a negative lens, the fourth lens is a meniscus positive lens, and the fifth lens is a negative lens. The fifth lens is a positive lens, and the first lens and the second lens are cemented lenses. According to the invention, through combination of a plurality of lenses and the balsaming lens, high optical index, good temperature adaptability, low distortion, large target surface and large entrance pupil diameter are realized; by reasonably distributing the focal power of each lens and adopting the combination of optical materials with specific refractive index and Abbe number, various aberrations such as spherical aberration, curvature of field, chromatic aberration and the like are effectively corrected, so that high-quality imaging with system RMS wavefront less than 0.02 lambda is realized, the distortion is strictly controlled below 0.02%, and the imaging quality is greatly improved. And a sub-pixel-level accurate positioning capability is realized for high-precision machine vision and measurement detection.

The application discloses a 4K thermal stable DVR optical imaging system, which comprises a first lens, a second lens, a third lens, a diaphragm, a fourth lens, a fifth lens, a sixth lens, a seventh lens, a filter, a protective glass and an image plane arranged in the direction from the object side to the image side. Through the reasonable use of the lenses with specific shapes and the definition of the optical power of each lens, the application realizes a large target surface design, the maximum target surface is greater than or equal to 9.56 mm, can match a 1 / 1.8 chip IMX728, a large aperture design F1.8, and can realize the recording and identification of objects on both sides of the road in a dark environment; the FOV is increased to 171 degrees, a large-angle picture can be shot, and the field of view is improved.

The embodiment of the invention discloses a line scanning industrial lens. The line scanning industrial lens comprises fourteen lenses which are sequentially arranged from the object space to the image space along the optical axis. Wherein the first lens to the seventh lens form a fixed group, the eighth lens to the twelfth lens form two movable groups, the thirteenth lens and the fourteenth lens form three fixed groups, and the two movable groups move back and forth along an optical axis to realize object distance change of the line scanning industrial lens; according to the line scanning industrial lens provided by the embodiment of the invention, through matching of lens materials, reasonable distribution of focal power of each lens and adoption of a floating focusing mode, the design of the line scanning lens capable of giving consideration to low distortion, large target surface and high resolution within a working distance of 200mm-1000mm is realized, the focal length of the system is 31mm, the maximum aperture is F4, the optimal object distance is 600mm, and the distortion lt is realized; 1%, and the imaging quality can reach MTF71pl / mmgt; and 0.3.

This invention discloses a wide-band, large-area infrared imaging lens covering short-wave and mid-wave bands, comprising a first lens, a second lens, a third lens, a fourth lens, a fifth lens, a sixth lens, and a seventh lens arranged sequentially from the object plane to the image plane; the first lens is a biconvex lens with positive refractive power; the second lens is a biconcave lens with negative refractive power; the third lens is a plano-concave lens with negative refractive power; the fourth lens is a convex-concave lens with positive refractive power; the fifth lens is a convex-concave lens with negative refractive power; the sixth lens is a biconvex lens with positive refractive power; and the seventh lens is a convex-plano lens with positive refractive power. This wide-band, large-area infrared imaging lens covering short-wave and mid-wave bands is suitable for the 1.5-5.5µm band, can be matched with high-resolution 1280x1024 (10µm) infrared detectors, has high optical throughput, can be used for imaging objects at greater distances, and has a compact structure.

The application discloses an ISAR target recognition method, system and device based on deep learning, relates to the technical field of radar target recognition, and comprises the following steps: firstly, a database is constructed through FEKO simulation, and is processed through labeling and data enhancement; secondly, a convolutional neural network with residual connection is built and pre-trained; thirdly, a three-scale feature pyramid YOLOV3 detection network is constructed, and anchor points are clustered by using k-means; then, joint training and optimization are carried out; finally, two types of detection systems are tested and generated, and target recognition is realized. The ISAR target recognition method, system and device based on deep learning adopt a feature pyramid network and k-means clustering anchor points, design nine groups of special anchor boxes according to the width-height distribution characteristics of ISAR targets, combine an up-sampling feature fusion strategy, realize effective detection of multi-pixel scale targets, and compared with a traditional YOLOV3 fixed anchor point scheme, the recall rate of cross-scale targets is significantly improved, and the industry pain points of small target missed detection and large target boundary blur are solved.

The invention provides a target detection and map construction device and method, equipment and a medium, and the method comprises the steps: carrying out the feature extraction of perception data, obtaining a multi-scale feature, determining the initial position and the target size of a target object based on the perception data, determining a target scale feature matched with the target object based on the target size and the multi-scale feature, and carrying out the detection of a map. And based on the initial position and the target scale feature, determining a local feature of the target object, and based on the local feature and the initial position, determining a target detection result of the target object. Therefore, the target scale features are dynamically determined through the target size, the features with more matched scales can be dynamically selected for the targets with different sizes, the problem that details of small targets are lost or redundant information of large targets is interfered by fixed single-scale features is avoided, the pertinence and effectiveness of feature expression are enhanced, and the accuracy of feature expression is improved. Therefore, the accuracy of target detection can be improved.