Patents
Literature
Patsnap Eureka AI that helps you search prior art, draft patents, and assess FTO risks, powered by patent and scientific literature data.

522 results about "Body images" patented technology

Generating ground truths for generative ai applications

A first neural network is trained to generate a ground truth using a small set of example images that illustrate the goal ground truth output images, which can be full-body images of people in an AR style. The first neural network is used to generate ground truth output images from random input images. Example methods of the first neural network include determining poses in input images, changing values of pixels within areas of the input images, inputting the poses, the areas of the changed input images, and a text prompt describing the input images, into a neural network, to generate output images. The methods further include determining losses between the output images and the input images and updating weights of the neural network based on the losses. A second neural network is then trained using the generated ground truth. And, an application is generated that uses the second neural network.
Owner:SNAP INC

Electric tricycle body spraying defect detection method and system based on machine vision

The invention provides an electro-tricycle body spraying defect detection method and system based on machine vision, and relates to the technical field of defect detection, and the method comprises the steps: obtaining a machine vision perception module; performing multi-angle image acquisition through a panoramic lens; performing preprocessing and panoramic stitching on the multi-angle tricycle body image set; performing defect area identification and anchor frame marking on the panoramic image of the tricycle body; carrying out feature extraction quantization on the tricycle defect area anchor frame set by adopting a microscopic lens; and carrying out statistical evaluation analysis on the tricycle defect characteristic quantitative information according to the spraying defect index set to generate an electro-tricycle body spraying defect detection result. The technical problem that in the prior art, due to the fact that small defects are difficult to accurately position, the defect detection accuracy is poor can be solved, comprehensive detection of the spraying defects of the electro-tricycle body is achieved by integrating the panoramic lens and the microscopic lens, and the defect detection accuracy is improved.
Owner:XUZHOU TIANYING AUTOMOBILE CO LTD

Hydraulic engineering dam body crack detection method and system based on machine vision

The invention discloses a hydraulic engineering dam body crack detection method and system based on machine vision, and relates to the technical field of computer vision. A high-definition camera is used for shooting a dam body image, obtaining a sample data set, extracting feature parameters of different noise types, and constructing a feature database; filtering the current dam body image to obtain a first dam body image, enhancing the crack gray scale difference of the first dam body image, obtaining a second dam body image, segmenting the second dam body image, screening candidate cracks in the crack image, and identifying real cracks according to the edge features of the candidate cracks. According to the method, a feature database is constructed, a targeted dam body image processing flow is combined, a high-definition camera collection and automatic processing flow is used, filtering parameters are dynamically adjusted by constructing the feature database, gray level enhancement and precise segmentation are combined, real cracks are precisely recognized through edge feature analysis and comparison, and quantitative parameters are output; and a reliable basis is provided for dam body safety assessment.
Owner:BOSHI INTELLIGENT TECH (CHONGQING) CO LTD

Vehicle body paint surface defect dynamic detection method, device, system and equipment

The invention relates to a vehicle body paint surface defect dynamic detection method, device, system and equipment, and the method comprises the steps: obtaining a first vehicle body image collected by a fixed support image collection assembly, and a second vehicle body image collected by a mechanical arm handheld image collection assembly; identifying a first vehicle body paint surface defect result in the first vehicle body image and a second vehicle body paint surface defect result in the second vehicle body image; joint calibration is conducted on a camera in the fixed support image acquisition assembly and a camera in the mechanical arm handheld mode image acquisition assembly, and a camera joint calibration result is obtained; and based on a camera joint calibration result, mapping the first vehicle body paint surface defect result and the second vehicle body paint surface defect result into a vehicle body three-dimensional coordinate system to obtain a vehicle body paint surface defect detection result. In the whole process, paint surface defect detection in a fixed support and mechanical arm handheld mode is compatible, and dynamic and accurate vehicle body paint surface defect detection can be achieved without keeping the vehicle body relatively static.
Owner:SPEEDBOT ROBOTICS CO LTD

Unmanned aerial vehicle bridge crack beam disease identification method based on progressive domain adaptation strategy

The invention discloses an unmanned aerial vehicle bridge crack beam disease identification method based on a progressive domain adaptation strategy. A bridge image is acquired through an unmanned aerial vehicle, the bridge image acquired by the unmanned aerial vehicle is subjected to overlapping type blocking, the segmented bridge image is marked, and bridge body image blocks are classified by adopting a CNN model; performing small-scale marking on cracks in the obtained bridge body image blocks, mixing the cracks with a public data set to obtain a mixed data set, and training a crack segmentation model based on the mixed data set; and carrying out type classification on the binary image after crack segmentation, and measuring a crack index at the same time. The bridge crack disease detection problem is decomposed into three stages of bridge body image block classification, bridge body crack segmentation, crack classification and geometric morphology estimation. The domain adaptation problem is dispersed to three different stages, so that the labeling cost is reduced, and the pressure of data acquisition is reduced. And an effective solution is provided for the bridge crack disease problem.
Owner:YICHANG CITY BRIDGE CONSTR INVESTMENT CO LTD

Train running state evaluation system based on data analysis

The invention relates to the field of train informatization, in particular to a train running state evaluation system based on data analysis. The train braking stress diffusion abnormal tendency value is determined by setting the braking information responding to the train, calling the train body image of the corresponding braking time period, extracting the train shaking characteristics and combining the discrete difference of the loads of all the carriages in the braking time period, and then the data time period type is marked. And the data in the time period is classified, uploaded and stored according to different marks, so that the real-time bandwidth requirement is reduced. According to the invention, the inter-wheel temperature difference characteristic and the in-wheel temperature difference characteristic of each wheel are considered and extracted, the wheel end abnormal risk characterization value is calculated and compared with the early warning threshold value, and the wheel end data is analyzed in a targeted manner, so that the data processing efficiency of a large amount of train data is improved, the real-time communication bandwidth demand is reduced, and the identification accuracy is improved.
Owner:BEIJING MASS TRANSIT RAILWAY OPERATION CORPORATION LIMITED

Tracking system and method

A tracking system is provided. A head-mounted display device is adapted to be mounted on a head of a user and comprises a camera and a processor. The camera obtains a body image includes a body portion of the user. The processor is configured to: determine a relative position between a body motion sensor of a body tracker and the body portion of the user based on the body image; and perform a target tracking of the body portion based on the relative position and sensor data of the body motion sensor of the body tracker. A body tracker is adapted to be mounted on the body portion of the user and comprises a body motion sensor. The body motion sensor obtains the sensor data. The sensor data indicates a movement and / or a rotation of the body portion.
Owner:HTC CORP

Auxiliary rehabilitation teaching system and method based on somatosensory interaction

The invention provides an auxiliary rehabilitation teaching system and method based on somatosensory interaction, and the method comprises the steps: determining an initial motion track of a target user during somatosensory interaction, capturing a body image of the target user during somatosensory interaction through a depth camera, determining skeleton points with the positions changed in the body image, and obtaining the skeleton points with the positions changed in the body image; performing drift compensation on the initial movement track based on the dynamic drift features of all skeleton points to obtain an actual interaction movement track when the target user performs somatosensory interaction; based on a standard rehabilitation action matched with the rehabilitation training action when the target user performs somatosensory interaction, performing deviation recognition on the interaction motion track to obtain a posture deviation degree between the rehabilitation training action and the standard rehabilitation action; and generating a feedback adjustment instruction of the target user in the somatosensory interaction process based on the posture deviation degree. By adopting the scheme of the invention, the drift error of inertial sensing can be corrected through the depth camera in the somatosensory interaction process so as to improve the motion capture precision of the auxiliary rehabilitation teaching system.
Owner:XIN HUA HOSPITAL AFFILIATED TO SHANGHAI JIAO TONG UNIV SCHOOL OF MEDICINE

Driver fatigue state detection system based on computer vision processing

The invention discloses a driver fatigue state detection system based on computer vision processing. The system comprises a facial feature extraction unit which selects a high-definition camera to capture a driver facial image; the upper body feature monitoring unit collects upper body image data through a high-definition camera to accurately extract the action frequency of the upper body; the hand-held steering wheel characteristic monitoring unit is used for sensing pressure applied by hands of a driver by arranging a grip strength sensor on a steering wheel through a protective sleeve; the data processing unit receives data information collected by the facial feature extraction unit, the upper body feature monitoring unit and the handheld steering wheel feature monitoring unit for feature fusion; the fatigue state judgment unit judges the fatigue state of the driver according to a preset fatigue judgment rule; the face blinking frequency is combined with multi-dimensional information such as the upper body action frequency and the grip strength, the state of a driver can be reflected from different angles, richer and more comprehensive data support is provided for fatigue judgment, and the fatigue state can be judged more accurately.
Owner:SHANDONG JIANZHU UNIV

Eye and hand tracking utilizing lensless camera and machine learning

Eye and hand tracking systems in head-mounted display (HMD) devices are arranged with lensless camera systems using optical masks as encoding elements that apply convolutions to optical images of body parts (e.g., eyes or hands) of HMD device users. The convolved body images are scrambled or coded representations that are captured by a sensor in the system, but are not human-recognizable. A machine learning system such as a neural network is configured to extract body features directly from the coded representation without performance of deconvolutions conventionally utilized to reconstruct the original body images in human-recognizable form. The extracted body features are utilized by the respective eye or hand tracking systems to output relevant tracking data for the user's eyes or hands which may be utilized by the HMD device to support various applications and user experiences. The lensless camera and machine learning system are jointly optimizable on an end-to-end basis.
Owner:MICROSOFT TECHNOLOGY LICENSING LLC

Vehicle body paint surface defect detection method and device, computer equipment and medium

The invention relates to a vehicle body paint surface defect detection method and device, computer equipment, a storage medium and a computer program product, and the method comprises the steps: obtaining a vehicle body digital model; obtaining a relative position relation between the defect detection camera and the vehicle body and a relative position relation between the production line and the vehicle body; performing paint surface defect detection according to a production line vehicle body image acquired by the defect detection camera to obtain a paint surface defect position under a camera coordinate system; based on the relative position relation between the defect detection camera and the vehicle body, the paint surface defect position under the camera coordinate system is positioned to the vehicle body digital model, and the paint surface defect position under the vehicle body coordinate system is obtained; and according to the relative position relationship between the production line and the vehicle body, mapping the paint surface defect position under the vehicle body coordinate system to the production line vehicle body to obtain a paint surface defect detection result of the production line vehicle body.
Owner:SPEEDBOT ROBOTICS CO LTD

Fish tank dynamic interaction display device based on position detection and interaction implementation method

The invention discloses a fish tank dynamic interaction display device based on position detection and an interaction implementation method. Belongs to the technical field of intelligent aquarium equipment and digital interaction and collects fish body images in real time, a control screen supports touch and voice interaction, and a background display screen dynamically updates display content according to the position of a fish body. The control processing hardware module integrates a microprocessor, a real-time digital graphic processing chip and an artificial intelligence processing unit, and image correction, multi-fish body detection tracking, coordinate mapping and dynamic background rendering are achieved. The method comprises the steps of image acquisition, geometric correction, position detection, coordinate mapping, dynamic rendering, cross-device linkage and the like, and supports gesture interaction, AI background generation and multi-fish tank linkage functions. Through combination of real-time position detection and a dynamic display technology, immersive interactive experience of fish moving and scene following is constructed, the problem that a traditional fish tank is insufficient in interactivity is solved, and the fish tank is suitable for families, commercial display and other scenes.
Owner:COMMUNICATION UNIVERSITY OF CHINA

Method and device for acquiring seismic diffraction body

The invention belongs to the technical field of seismic data acquisition and processing, and particularly relates to a method and a device for acquiring a seismic diffraction body. The method comprises the following specific steps: preprocessing collected seismic data to obtain seismic shot gather data which can be directly used for migration; eliminating residual noise by using a multi-channel singular spectrum analysis technology to obtain optimized pre-stack gather data; a complete diffracted wave is obtained through forward and reverse transformation of a virtual seismic source and through offset-reflected wave removal-deoffset; and finally, accurately separating weak signals of the diffracted waves under the condition of intersection and tangency of seismic wave fields by researching characteristics that the polarity and the amplitude of the diffracted waves change along with an emergence angle and an azimuth angle, thereby realizing high-resolution diffraction body imaging of the fractured-vuggy reservoir, effectively improving the imaging precision of the karst fractured-vuggy reservoir, and achieving the purpose of fine detection of the fractured-vuggy reservoir. The method is of positive significance to clarify the seismic response of the fracture-vug type reservoir, implement fracture-vug distribution characteristics and guide effective identification of the karst type reservoir and optimization of a favorable target area.
Owner:CHINA PETROLEUM & CHEMICAL CORP +1

Driver fatigue monitoring method and device, computer equipment and medium

The embodiment of the invention provides a driver fatigue monitoring method, which relates to a driver fatigue technology, and comprises the following steps: aiming at each frame of an upper body image of a driver, judging an eye state and a mouth state of the frame of image; extracting high-level features of a plurality of RO I areas of each frame of image, and fusing the high-level features to obtain combined features; performing feature feedback prediction to obtain an initial fatigue state; determining an initial eye state and an initial mouth state; and carrying out weighted integration on the states to obtain a final fatigue estimation result. According to the technical scheme, firstly, a plurality of RO I regions are introduced to obtain combined features with higher characterization capability, in addition, an existing sequential network structure is improved, and a more effective fatigue probability prediction value is obtained through feature feedback; and the fatigue prediction value after weighted synthesis is obtained by combining the fatigue prediction value of the key point detection and the fatigue prediction value of the sequential network, so that the fatigue prediction result is more accurate, and the problems of false report and missing report in the current fatigue detection technology can be effectively solved.
Owner:WUHAN JIMU INTELLIGENT TECH CO LTD

Defect detection method and device based on dynamic double-threshold template and storage medium

The invention discloses a defect detection method and device based on a dynamic dual-threshold template, and a storage medium, and aims to improve the defect detection precision of an image shot by a display screen. Collecting a plurality of batches of defect-free sample images of the target model display screen; counting gray extreme value boundary data of each pixel point on the plurality of batches of defect-free sample images; and generating a lower limit standard template and an upper limit standard template according to the gray extreme value boundary data. And obtaining a regional difference amplification coefficient and a compensation offset. And generating a dynamic detection threshold template for the lower limit standard template and the upper limit standard template through the regional difference amplification coefficient and the compensation offset. And obtaining a to-be-tested screen body image of the target model display screen. And carrying out pixel-level difference on the to-be-detected screen body image and the dynamic detection threshold template to generate a defect mask image. And performing defect detection on the display screen through the defect mask image to generate a first defect detection result.
Owner:SHENZHEN SEICHITECH TECHN CO LTD

Portrait-based controllable style migration and image art enhancement method and system

The invention relates to the technical field of image style migration, in particular to a portrait-based controllable style migration and image art enhancement method and system, and the method comprises the steps: carrying out the preprocessing of an obtained content image, enhancing the contrast and definition of the image, and obtaining a preprocessed image; based on a preset image segmentation model, pixel-level semantic segmentation is carried out on the preprocessed image, and a figure main body image is output; obtaining style information, based on a preset style generation model, performing style migration on the character main body image according to the style information, and generating a stylized character image; obtaining a target style image, and fusing the figure image and the target style image based on a preset ComfyUI workflow to obtain a final processing image; the image processing method and the image processing device have the advantages that on the premise of high customization flexibility, the quality of the image after style migration is improved, and the coordination between the portrait main body and the background is improved.
Owner:HUASHU (ZHEJIANG) TECH CO LTD

Vehicle body type distinguishing method and device based on intelligent camera and electronic equipment

The invention discloses a vehicle body type distinguishing method and device based on an intelligent camera and electronic equipment, and relates to the field of data processing. The method comprises the following steps: acquiring a vehicle body image; segmenting a vehicle body area in each vehicle body image to obtain vehicle body features, and establishing a multi-dimensional feature vector; if it is determined that the similarity between the multi-dimensional feature vector and the target image feature is greater than a preset similarity threshold, determining a target vehicle body type corresponding to the to-be-identified vehicle body as a vehicle body type corresponding to the target image feature; determining a plurality of derived types corresponding to the target vehicle body type and corresponding image feature subsets; performing secondary segmentation on the vehicle body image according to the image feature subset to obtain corresponding local features; matching the local feature with each image feature subset to obtain a target derived category; and according to the target vehicle body type and the target derived type, adjusting production parameters corresponding to the to-be-identified vehicle body. By implementing the technical scheme provided by the invention, the accuracy of vehicle body type identification on the production line is improved.
Owner:BEIJING HINSONGYICHANG MACHINERY & ELECTRIC ENG

Realistic three-dimensional reconstruction method for earth surface exposed geologic body based on neural radiation field

The invention discloses an earth surface exposed geologic body reality sense three-dimensional reconstruction method based on a neural radiation field. The method comprises the steps that an earth surface geologic body image data set is acquired; performing feature matching on the image data set by using a feature matching strategy based on a nearest neighbor algorithm and a LightGlue neural network to generate feature point matching information and a feature description file; sparse three-dimensional point cloud data and camera pose information are obtained based on incremental SfM; and obtaining depth information of an image data set based on sparse three-dimensional point cloud data and camera pose information, and establishing a geologic body realistic three-dimensional reconstruction model by using a neural radiation field optimization method fusing the depth information and Gaussian position coding. According to the method, the feature matching precision and the camera pose calculation accuracy are improved, the texture detail expression ability of a geological structure can be improved, the reconstruction quality of a surface geologic body model is enhanced, and geological information support can be provided for the fields of geological research and engineering application.
Owner:CHINA UNIV OF GEOSCIENCES (WUHAN) +2

Text-based multi-modal face generation method and device, equipment and storage medium

The invention provides a text-based multi-modal face generation method and device, equipment and a storage medium, and belongs to the technical field of face image generation, and the method comprises the steps: determining a main body image based on a reference image and a main body mask corresponding to the reference image, and determining a decoupling global identity embedding feature based on the main body image. The reference image is an initial face image. And determining a multi-modal local identity embedding feature based on the reference image and the mask image corresponding to the reference image. The multi-modal local identity embedding features are text embedding class features. And determining a target generation face image based on the decoupled global identity embedding feature and the multi-modal local identity embedding feature. According to the text-based multi-modal face generation method and device, the equipment and the storage medium provided by the invention, the accuracy of text-to-face generation can be improved, and actual requirements are met.
Owner:BEIJING UNIV OF CIVIL ENG & ARCHITECTURE

Body-in-white autonomous welding method based on machine vision

The invention discloses a body-in-white autonomous welding method based on machine vision, relates to the field of automobile manufacturing and body-in-white welding, and solves the problems that the welding quality is influenced due to the fact that an existing welding robot depends on welding conditions, the flexibility is poor, the working efficiency of welding is reduced, the production cycle of a product is prolonged, and finally the production cost is increased. According to the method, the camera is calibrated, and the internal reference and the distortion coefficient of the camera are obtained; after obtaining a vehicle body image, extracting two-dimensional coordinates of centroids of feature points; according to the mass center two-dimensional coordinates and the feature point mass center three-dimensional coordinates of the vehicle body model, the coordinate conversion relation from a vehicle body coordinate system to a camera coordinate system is solved; hand-eye calibration is conducted, and the coordinate conversion relation from a camera coordinate system to a mechanical arm coordinate system is calibrated; the position coordinates of the welding spots are converted into a mechanical arm coordinate system from a vehicle body coordinate system, and a mechanical arm conducts welding. According to the method, the positions of all the welding spots do not need to be recognized, the calculated amount is reduced, and the requirement for hardware is lowered accordingly.
Owner:JILIN UNIVERSITY

Shale pore type identification and quantitative characterization method based on digital core

The invention provides a shale pore type identification method based on a digital core, and the method comprises the steps: carrying out the gray adjustment processing and image noise reduction processing of an original gray image of a to-be-identified shale sample, and obtaining a standard gray image; organic matter and transition zone images in the standard grayscale image are extracted, organic matter identification processing is carried out, and an organic matter connected body image is determined; extracting a pore image in the standard gray-scale image, performing pore recognition processing in combination with the organic matter connected body image, and determining an organic pore image and an inorganic pore image; and performing micro-crack identification processing on the inorganic pore image, and determining a micro-crack image and an inorganic pore image. By applying the method, the organic pores, the inorganic pores and the microcracks of the shale can be quantitatively identified, the pore structure of the shale can be quantitatively represented, and the structural parameters of the organic pores, the inorganic pores and the microcracks can be analyzed.
Owner:CHINA PETROLEUM & CHEMICAL CORP +1

Large model-based large yellow croaker growth stage feeding amount evaluation method and system

The invention discloses a large model-based large yellow croaker growth stage feeding amount evaluation method and system, and belongs to the field of aquaculture, and the method comprises the steps: collecting underwater images of a fish pond in a plurality of shooting directions at the same time, and recognizing and segmenting a fish body image corresponding to each entity fish in each underwater image; determining the distance between an entity fish corresponding to the fish body image and a camera for collecting the fish body image, and extracting fish body appearance features of the fish body image; randomly selecting a fish body image from each shooting direction to form a fish body image association combination, and screening a second fish body image association combination from all possible fish body image association combinations according to the distance and fish body appearance features; and constructing three-dimensional point cloud data of the entity fishes according to the second fish body image association combination, fitting three-dimensional body shape parameters of the entity fishes through the three-dimensional point cloud data, and determining the final feeding amount of the fish school according to the three-dimensional body shape parameters of the entity fishes. According to the invention, the problem of low accuracy of the real-time feeding amount of the fish school can be solved.
Owner:GUANGDONG OCEAN UNIVERSITY

Flounder fish growth discrimination system and method

The invention relates to the technical field of aquaculture. The invention discloses a flounder fish growth discrimination system and method, and the method comprises the steps: employing a high-definition camera to carry out the multi-angle shooting of a cultured fish body, obtaining a non-shielding, non-overlapping and high-resolution cultured fish body image, building a YOLOv8 segmentation model, building a recognition module, enabling the recognition module to obtain a measurable parameter and a frame parameter based on a generated mask of the fish body, and carrying out the recognition of the measurable parameter and the frame parameter. After data correction is completed, variety analysis and identification are completed, variety information is input into the YOLOv8 model, and the growth condition of cultured fish bodies is judged according to the final growth conditions of the body thickness, the body length and the body width. According to the method, cultured fish body image segmentation is completed by establishing the YOLOv8 segmentation model, and after measurable parameters and frame parameters are obtained, automatic generation and discrimination are completed according to body thickness derivation results of different varieties, so that the efficiency and precision of culture management are improved, and the method is suitable for large-scale popularization and application. Powerful data support is provided for scientific feeding, feed optimization, growth regulation and control, disease early warning and the like in the breeding process.
Owner:YELLOW SEA FISHERIES RES INST CHINESE ACAD OF FISHERIES SCI

Screen locking image display method, electronic equipment and readable storage medium

The invention discloses a screen locking image display method, electronic equipment and a readable storage medium, and belongs to the technical field of terminals. The method is applied to the electronic equipment and comprises the steps that in response to a screen turn-on operation, the electronic equipment displays a first screen locking interface, and the first screen locking interface comprises a first image, a system time indicator displayed on the first image and a main body image generating partial shielding on the system time indicator; in response to the rotation operation on the electronic equipment, the electronic equipment displays a second screen locking interface, and the second screen locking interface comprises a second image, a system time indicator displayed on the second image and a main body image which generates partial shielding on the system time indicator. The relative position between the main body image and the system time indicator is not changed, so that the depth-of-field image display effect before and after rotation of the display screen is ensured, and the depth-of-field image display attractiveness is improved.
Owner:HONOR DEVICE CO LTD

Motion function data acquisition method and system based on human-computer interaction

The invention relates to a motion function data acquisition method and system based on human-computer interaction. The motion function analysis method is applied to a motion function analysis device and comprises the steps that in response to an initial position instruction of an abnormal object, initial motion analysis is conducted based on historical treatment data of the abnormal object, and initial motion data is presented; when the corresponding initial body image is detected, pose analysis is executed, and a body condition labeling box is displayed to a labeling person. After annotation confirmation is received, carrying out movement analysis in combination with annotation data and historical treatment data, and outputting action change data; and if the current body image corresponding to the current body image is detected, taking the current body image as a new initial image, and repeating the pose analysis and labeling process. And until an annotation stop instruction is received, the system summarizes each annotation and historical treatment data to carry out object function analysis, and motion function data of the abnormal object is generated. By adopting the method, the accuracy of understanding the current abnormal information of the abnormal personnel can be effectively improved.
Owner:BEIJING TIANTAN HOSPITAL AFFILIATED TO CAPITAL MEDICAL UNIV

Glass bottle defect detection method

The invention discloses a glass bottle defect detection method, which belongs to the technical field of defect detection, and comprises the following steps: labeling defect areas for all images and adding corresponding defect labels to obtain a first preprocessed image with the defect areas and the defect labels; respectively adopting corresponding pre-training models for a first pre-processed image in the bottle body image library, a first pre-processed image in the bottle opening image library and a first pre-processed image in the bottle bottom image library, taking the first pre-processed images as input, and outputting a predicted defect area and a predicted defect label, incremental learning is carried out on the corresponding pre-training models, and detection models corresponding to the bottle body, the bottle opening and the bottle bottom are obtained respectively; and respectively inputting the newly obtained images into the corresponding detection models to obtain corresponding defect labels, defect areas and confidence coefficients. The glass bottle defect detection method solves the problem that at present, different parts need to be detected at different times, and rapid detection of the glass bottle is difficult to achieve.
Owner:FOSHAN JINGHUA VISION TECH CO LTD

Vehicle body paint surface defect detection method and device, computer equipment and medium

The invention relates to a vehicle body paint surface defect detection method and device, computer equipment, a storage medium and a computer program product, and the method comprises the steps: obtaining an initial vehicle body image, collected by a streak camera, of a vehicle body under the irradiation of sine stripe light; performing defect enhancement on the initial vehicle body image by adopting an optical flow method to obtain a defect-enhanced vehicle body image; performing defect detection on the defect-enhanced vehicle body image to obtain a vehicle body paint defect position under a camera coordinate system; and mapping the vehicle body paint surface defect position under the camera coordinate system into a vehicle body three-dimensional coordinate system to obtain a target vehicle body paint surface defect position. In the whole process, the streak camera collects the initial vehicle body image of the vehicle body under the irradiation of the sine stripe light, vehicle body paint surface defect detection is performed based on the initial vehicle body image, and the optical flow method is adopted to perform defect enhancement on the initial vehicle body image, so that efficient and accurate vehicle body paint surface defect detection can be realized.
Owner:SPEEDBOT ROBOTICS CO LTD

Commodity marketing picture and video generation method and device, equipment and medium

The invention discloses a commodity marketing picture and video generation method and device, equipment and a medium. The method comprises the following steps: receiving an input commodity image and description information provided by a user; carrying out preprocessing and main body segmentation on the commodity image to obtain a commodity main body image; based on the commodity main body image and the user description information, commodity content understanding is carried out through a multi-modal model, a structured commodity semantic tag is generated, and the commodity semantic tag comprises the category, the visual feature and the key attribute of the commodity; and on the basis of the commodity semantic label, creative content generation and automatic typesetting processing are carried out through a multi-modal model, and a commodity marketing image containing a background, an advertisement copywriting and decoration elements is generated. According to the method, the tedious process needing multiple stages, multiple tools and repeated manual intervention is integrated into a highly-automatic end-to-end process, the manufacturing efficiency of commodity marketing materials is improved, meanwhile, the requirement for professional skills of a user is lowered, and the consistency of the visual context, style and brand is guaranteed.
Owner:BEIJING WEZONET NETWORK TECH CO LTD

Detection and identification system

A detection and identification system includes a display portion, an image-taking portion, and a protective clothing detection portion. The image-taking portion captures a full-body image of a person wearing protective clothing. The display portion displays the full-body image in real time. The protective clothing detection portion divides the full-body image into plural to-be-identified regions according to the joints of the human body, uses an anomaly detection method or semantic segmentation algorithm to determine whether each to-be-identified region meets detection criteria, and marks the to-be-identified regions in such a way that each to-be-identified region meeting, or failing to meet, the detection criteria is displayed by the display portion as marked with a first mark or a second mark respectively. The first and second marks are different-colored blocks or frames. The system allows the current condition of the protective clothing to be rapidly known so that appropriate handling measures can be taken.
Owner:FENG CHIA UNIVERSITY

Tuna quality grading intelligent detection method based on multivariate information feature fusion

The invention discloses a tuna quality grading intelligent detection method based on multivariate information feature fusion, and relates to the technical field of fish quality grading detection. The method comprises the following steps: acquiring fish body images and meat quality images of a plurality of different tunas, collecting related text information of the tunas, including producing areas, varieties, weights and body lengths, and constructing a data set; preprocessing the image and the related text information; generating a multi-modal fusion feature vector through feature extraction and fusion; inputting into a tuna quality grade detection module; and training the deep learning model, and subsequently performing quality grading detection on the to-be-detected tuna. According to the method, the fish body image, the meat quality image and the related text information are fully considered, the multi-modal fusion features including the fish body, the meat quality and the related text information are generated by adopting the multi-modal feature fusion technology, and training is performed through the quality grading detector, so that the robustness and the accuracy of detection are improved, and the detection result is more comprehensive and reliable.
Owner:CHINESE ACAD OF FISHERY SCI