542 results about "Video detection" patented technology

The invention discloses an AI generation content detection and review method, device and equipment and a storage medium, and the method comprises the steps: receiving multi-modal input data which comprises text data, image data and video data; calling a large language model to carry out compliance analysis on the text data to obtain an analysis result; when the analysis result is compliance, calling a multi-modal model to identify and detect the image data and the video data, and respectively obtaining an image detection result and a video detection result; and aggregating the analysis result, the image detection result and the video detection result to obtain a content detection result. According to the method, the processing paths are automatically allocated according to the content types, redundant calculation is avoided, hardware consumption is remarkably reduced, the multi-modal detection results are aggregated into structured output, the large-batch content processing efficiency is remarkably improved, calculation resource occupation is greatly reduced, and seamless integration of the detection results and a downstream service system is achieved.

The invention discloses a multi-mode counterfeit video detection method based on a multi-head addition cross attention mechanism, and belongs to the technical field of video counterfeit detection. The method comprises the following steps: preprocessing a video stream, decomposing a single-frame positioning face, and extracting an audio to generate a Mel spectrogram slice; the 3D convolutional network extracts video spatio-temporal features and motion differences, and the filter bank extracts audio features in combination with the residual network; audio features are mapped to a video alignment space through asymmetric projection, the video features are subjected to bidirectional interaction with an audio input multi-head addition cross attention module after being subjected to time sequence coding, and audio dominant and video dominant features are generated and are cascaded and fused with original features; and constructing cross-modal similarity loss constraint feature distribution, and fusing feature dynamic weighting and time sequence compression to output four classification probabilities of audio-visual double true, audio-visual double pseudo, video pseudo-audio true and video pseudo-audio pseudo. The multi-mode counterfeiting recognition precision is improved, and texture abnormity and audio and video mismatch features are captured.

The invention provides a video quality detection method based on multi-scene self-adaption. The method comprises the following steps: selecting an area in a video picture, extracting multi-dimensional features of the area, determining the type of a scene where the video picture is located, and dynamically adjusting video detection parameters of the video picture; detecting the quality of the video picture based on the video detection parameters; intercepting an abnormal video picture, and adding the abnormal video picture into the constructed equipment fault template library M; and performing local model retraining on the video quality detection model based on the equipment fault template library M, and detecting the quality of a to-be-detected video picture based on the video quality detection model. On the basis of deep analysis of different scene features, the scene type of the video picture is accurately judged through multi-dimensional feature calculation, and then the video quality detection parameters are dynamically adjusted according to the scene type, so that the detection system can automatically adapt to the optimal detection parameters according to scene changes, misjudgment and missed judgment caused by the scene changes are avoided, and the detection efficiency is improved. And the accuracy of video quality detection in different scenes is remarkably improved.

The invention relates to the technical field of video detection, in particular to a video anomaly detection method based on a large language model, and the method comprises the steps: constructing a searchable knowledge base through normal videos; after video clips are extracted from a to-be-detected video, a description text is obtained through the large language model; encoding the description text into a vector, and retrieving in a knowledge base to obtain a plurality of most similar behavior patterns; and inputting the most similar behavior patterns and the description text into the large language model, and reasoning whether the video to be detected is abnormal or not through guidance of a third prompt word. By introducing the large language model, the problem that a traditional video anomaly detection method is lack of interpretation is solved, a detailed natural language reasoning process can be output while anomaly detection is carried out, the background, behavior logic and reasoning basis of abnormal event occurrence are clearly described, and the understandability and decision reference value of a detection result are greatly improved.

The invention discloses an unsupervised deeply-forged face video detection method based on strong enhancement contrast learning, which comprises the following steps: firstly, preprocessing face video data, cutting out a full-size label-free image and a face region image, then preliminarily classifying the label-free data and endowing the label-free data with pseudo labels by using natural physiological characteristics in face features; a strong enhancement contrast learning strategy is adopted, a plurality of basic enhancement operations are randomly combined to generate a strong enhancement view with larger structure change and semantic difference, strong enhancement contrast learning of the face image is completed, and finally accurate detection of a counterfeit face video is realized through diclustering and authenticity discrimination. Compared with an existing unsupervised detection method, the method has the advantages that multi-level feature fusion and strong enhancement contrast learning are combined, deep counterfeit traces can be captured more effectively, higher robustness and generalization ability are shown in complex scenes such as noise and compression, and the method is high in robustness and generalization performance. And high-precision detection of the deeply-forged face video under an unsupervised condition is realized.

The invention belongs to the technical field of multimedia security, and particularly relates to a deep forgery detection method and system based on audio and video multi-mode fusion. The method comprises the following steps: extracting lip motion space-time characteristics from a video stream through dynamic ROI (Region of Interest) cutting, and processing an audio stream through fast Fourier transform and a Mel filter bank in sequence to obtain audio spectrum characteristics; reconstructing and generating a corresponding audio feature based on the spatio-temporal feature of lip motion, and fusing the audio spectrum feature and the generated audio feature by adopting a bidirectional cross-modal attention mechanism to obtain an attention fusion feature; a Mel spectrogram of the audio stream is acquired, and feature maps of different scales are extracted from the Mel spectrogram and the video stream by using a feature extraction model formed by a convolutional layer and are fused to obtain multi-scale features; after the attention fusion feature and the multi-scale fusion feature are flattened, performing weighted fusion after channel dimension splicing, and outputting a forgery probability through a multi-layer perceptron. The video detection performance is improved.

The invention discloses a railway freight car part replacement operation video detection system, solves the problem of how to improve the railway freight car maintenance operation detection efficiency, and belongs to the technical field of railway freight car maintenance. The method comprises the following steps: a video processing layer processes a vehicle depot operation video stream collected in real time to obtain a dynamic frame image; the multi-target detection and tracking module detects the dynamic frame image to obtain a state label of the target, determines the first occurrence time and the final disappearance time of the target and the position coordinate of the target, and forms time-space sequence data; the action time sequence analysis module is used for analyzing the state label and the space-time sequence data of the target, filtering out abnormal action nodes and extracting action nodes and corresponding time of a state conversion scene; the component state judgment module verifies the final setting state and the operation process compliance of the component according to the state label and the space-time sequence data of the target; and the decision output layer generates a structured detection report according to the analysis result.

The invention discloses a fake video detection method and system based on potential space learning, and belongs to the technical field of multimedia forensics, and the method comprises the steps: reinforcing the capturing capability of a model for the potential distribution difference of true and false videos through triple contrast learning, combining GRU network modeling time sequence dynamic features and 3D CNN extraction time-space content features, and obtaining a fake video detection result. The joint representation of video time-space domain features is realized, potential spatial features and content features are dynamically fused through an attention mechanism to focus and forge contradictory information, and a global self-attention mechanism of a time encoder is utilized to mine a long-range time sequence dependency relationship. The method overcomes the problems of insufficient generalization ability, weak time sequence modeling and single feature expression of an existing deep counterfeiting detection method, can effectively distinguish a high-quality counterfeiting video from a real video, improves the adaptability to an unknown counterfeiting technology, enhances the detection robustness of a time sequence anomaly and a complex counterfeiting scene, and improves the detection efficiency. And a solution is provided for deep counterfeiting identification.

The invention relates to a self-adaptive control method for a rain and snow mode of rail transit, electronic equipment and a medium, and the method comprises the steps: comprehensively studying and judging a rain condition based on multi-sensor data and numerical weather forecast data, and calculating a weather grade in the rain and snow mode, the multi-sensor data including video detection data, laser radar data and rainfall data; whether a rain and snow mode is set or not is judged according to the weather level, if yes, the current track slip degree is evaluated according to the multi-sensor data, and the GEBR value is adaptively updated based on the slip degree; the CC calculates the running safety braking envelope of the train by using the updated GEBR value, and feeds back a slip detection result and a dynamic parameter adjustment result in real time; and the ATS dynamically adjusts the operation plan of the train according to the slip detection result reported by the CC, and optimizes the operation scheduling of the whole train. Compared with the prior art, the method has the advantages that self-adaptive management of the rain and snow mode is achieved based on the dynamically-adjusted GEBR, and the running safety and reliability of the train in the rain and snow weather are effectively improved.

The invention relates to the field of urban drainage, and discloses an urban drainage pipe network operation safety assessment method and system, and the method comprises the steps: obtaining pipe network operation state data and pipe network physical structure data, and constructing a drainage pipe network GIS database; acquiring a pipeline defect detection image based on CCTV television detection and QV video detection technologies, and importing the pipeline defect detection image into a GIS system; performing model calibration dynamic simulation on the constructed drainage hydraulic model based on the data in the GIS and the real-time operation monitoring data of the drainage pipe network to obtain drainage pipe hydraulic data; performing defect evaluation on the sewage pipe network, the rainwater pipe network and the confluence pipe network by adopting different safety level models based on the drainage pipe hydraulic data; based on the corresponding defect probability, the risk is divided into a first-level risk, a second-level risk and a third-level risk; analyzing the pairwise variable relationship to find key indexes influencing the risk; and matching a corresponding coping strategy based on the key index influencing the risk, and displaying the coping strategy.

The invention discloses a deep counterfeit video detection method based on multi-dimensional feature collaborative modeling, which belongs to the field of computer vision and comprises the following steps of: preprocessing a complete face image by using a face recognition technology to obtain an image of a face local area; respectively extracting frequency domain features and spatial domain features of the face local region; carrying out adaptive weight modulation and nonlinear fusion on the extracted spatial domain features and frequency domain features by adopting a position sensing double-domain fusion module; the contribution degree of the local region in the local-global feature fusion process is adjusted, and a final global fusion feature is obtained; and the loss of each task is automatically weighted and balanced, and the prediction classification of the forged video is realized. According to the method, a multi-dimensional feature fusion framework of local and global and spatial and frequency domains is adopted, so that the detection precision of the deeply-forged video is effectively improved.

The invention provides a power transmission line icing detection method and system. The method comprises the following steps: collecting an icing image, temperature distribution data and sparse point cloud data; performing alignment processing on the sparse point cloud data to generate a distance mapping graph; inputting the icing image into a SegFormer model, and adjusting the expansion rate of a grouping depth separable convolutional layer through a distance mapping graph to obtain a first icing segmentation feature; generating an icing area pseudo label based on the temperature distribution data, and supervising the icing image to be segmented into a second icing segmentation feature through comparative learning; fusing and outputting the first icing segmentation feature and the second icing segmentation feature to obtain an icing feature; and according to the icing characteristics, extracting and complementing the corresponding sparse point cloud data, and constructing a three-dimensional model of power transmission line icing to obtain a power transmission line icing detection result, so that the technical problem of low accuracy of video detection of the icing condition of the power transmission line under the condition of severe weather in the prior art can be solved.

The invention provides a forged face video detection method and device based on multi-modal behavior consistency, electronic equipment, a storage medium and a program product. The method comprises the following steps: extracting voice features, facial expression features and head action features from a video signal to be detected; recognizing voice emotion, facial emotion and semantic emotion; based on the VAD value sequences of the various emotions, emotion consistency features and emotion synchronism features among the various emotions are calculated, and emotion semantic consistency features between the semantic content and the facial emotion and between the semantic content and the voice emotion are calculated; constructing a cross-modal time dependence graph to obtain interaction features; processing the voice features, the facial expression features and the head action features by using a hierarchical attention network to obtain time sequence features; a multi-dimensional fusion feature vector is formed; and processing the fusion feature vector by using a preset binary classifier to obtain a classification result indicating whether the to-be-detected video signal is a fake face video or not.

The invention belongs to the technical field of drainage pipeline defect detection, and particularly provides a drainage pipeline video defect detection method and system based on image quality screening, and the method comprises the steps: collecting and processing a detection image in a drainage pipeline, and constructing an image pair training set and a test set; constructing a GAN-based reference-free image quality initial evaluation model; training and testing the reference-free image quality initial evaluation model to obtain an image block quality evaluation discriminator; combining the image block quality evaluation discriminator with the image information entropy weight matrix to obtain a non-reference image quality final evaluation model; and carrying out quality screening on detection images in the drainage pipeline to be detected based on the non-reference image quality final evaluation model, and carrying out defect detection based on the trained YOLOv8 defect detection model to obtain a defect detection result. According to the invention, by screening out low-quality images, the burden of redundant image frames on subsequent defect detection tasks is effectively reduced, and the video detection speed is improved.

The invention relates to an AI generated video detection method and system fusing global and inter-frame semantics, and belongs to the technical field of artificial intelligence, computer vision and multimedia security. The method comprises the following steps: extracting advanced semantic features of a video frame by using a pre-trained CLIP model; modeling a video overall semantic distribution and narrative structure by adopting a Transform structure through a global semantic branch; capturing inter-frame local semantic mutation in parallel through an inter-frame semantic consistency branch by adopting BiLSTM and an attention mechanism; and finally, carrying out adaptive weighted fusion on discrimination results of the two branches through a dynamic fusion module, and outputting a final detection probability. According to the method, videos generated by multiple generation technologies such as GAN, a diffusion model and Sora can be effectively recognized from the high-level semantic dimension, and the method has high robustness to interference such as video compression and fuzziness and cross-model generalization ability and is suitable for actual scenes such as network content auditing and media evidence obtaining.

The invention relates to a video detection method and device for an edge algorithm all-in-one machine, equipment and a medium, and the method comprises the steps: obtaining task parameters corresponding to a polling scheduling task for each polling scheduling task; starting a plurality of detection algorithm instances corresponding to each algorithm set, and creating a detection queue for each detection algorithm instance; based on a global framing timetable corresponding to the plurality of polling scheduling tasks, performing framing on video streams collected by a plurality of video input devices corresponding to the plurality of to-be-detected video channel sets to obtain a plurality of video frames, and placing the plurality of video frames in respective corresponding detection queues; and for each detection algorithm instance, continuously extracting the video frame from the corresponding detection queue, and carrying out detection analysis on the extracted video frame. According to the invention, the adaptive matching of the frame taking operation and the detection requirement is realized, and the resource utilization rate and the detection real-time performance are improved under the limited computing power.

The invention discloses a face-changing video detection method and system based on illumination feature decoupling. The method comprises the following steps of: framing a video and extracting an RGB (Red, Green and Blue) image of a detection area; sending the RGB image into a learnable local gravity mode extraction module, and extracting a local gravity mode image; the RGB image and the local gravity mode image are sent to a double-branch high-semantic feature extraction module, and illumination related features and illumination invariant features are extracted respectively; the illumination related features are sent to a contrast learning module based on relighting; sending the classification features into a classifier to obtain prediction probability distribution, and carrying out dichotomy supervision by using cross entropy loss; training a model and storing the model; and the model test loading model outputs a detection result of the to-be-detected video. According to the method, the illumination features are decoupled by using the domain generalization strategy, interference of domain-related illumination information is effectively suppressed, the characterization capability of the detection features is improved, and the method has a relatively good detection effect and generalization capability.

The invention discloses a network live broadcast lagging detection and repair method based on dynamic threshold self-adaption, which belongs to the technical field of video detection, and comprises the following steps: S10, constructing a lagging detection index system for a client carrying out network live broadcast, the lagging detection index system comprising a network broadband index, a video buffer area index, a frame rate index and a packet loss rate index; and S20, network parameters of the client are obtained according to the constructed lagging detection index system, the network parameters comprise network topology structure information, network delay and network jitter, and the stability of the client in the current network environment is analyzed based on the obtained network parameters. According to the invention, by constructing the multi-dimensional lagging detection index system and the dynamic threshold adjustment mechanism and combining the lagging repair strategy, the smoothness and stability of network live broadcast are effectively improved, the user experience is optimized, and the lagging detection accuracy is improved.

The invention relates to the technical field of video detection, in particular to an anomaly detection method and system fusing large and small models and cooperating with cross-time domain perception. Comprising the steps of configuring a small model, a multi-modal large model and a cross-time domain parameter based on an electric power scene; acquiring a monitoring video stream, calculating a dynamic frame extraction frequency based on a scene change coefficient of continuous video frames of the monitoring video stream, extracting a frame sequence and recording a timestamp of each frame; inputting a frame sequence into the small model to obtain a candidate target, removing redundancy through non-maximum suppression, screening an effective target in combination with a single-frame confidence coefficient and an inter-frame time domain continuous overlapping rate, marking an alarm image and generating a preliminary analysis result; a time-domain-crossing video frame, power scene sensor data and environment data are collected according to a dynamic time window by taking an alarm image timestamp as a center, a visual time-domain feature and a power scene sensor time sequence feature are extracted, and a time-domain-crossing fusion feature is obtained through fusion. The problem of misjudgment and missed judgment caused by single-frame dependence in power scene anomaly detection is solved.

The invention discloses a speaking face generation video detection method and device based on multi-modal global and local collaboration, and the method comprises the steps: carrying out the framing and face region extraction preprocessing of a video, and enabling an input audio to be aligned with an input video frame in a time dimension; a speaking face generation video detection model is constructed and trained, an inconsistent region between video frames is focused on through a preset region attention smooth detection module and a difference capture-time frame aggregation module, and an original audio is input into an audio feature extraction module; carrying out consistency calculation on the input audio and video modal features through a preset audio and video fusion module, fusing the audio and video features so as to discriminate the forged generated video to obtain a video discrimination tag, and then calculating a loss training model; and finally, discriminating a to-be-detected video by using the trained model. According to the method, audio and video features are combined, so that the detection precision and robustness can be effectively improved.

The invention belongs to the field of computer vision, relates to a rockfall monitoring method and system based on visual identification, and aims to solve the technical problems that an existing scheme is difficult to achieve good balance among reliability, accuracy and economy and is poor in application effect. The method comprises the following steps: acquiring video stream data and vibration waveform data of a target area; monitoring according to the vibration waveform data, and when the vibration energy exceeds a preset first threshold value, generating a primary vibration event trigger and extracting video data slices from the video stream data; inputting the video data slices into a target detection network to identify candidate moving targets in the video frames and extract kinematic features; performing spectral analysis on the vibration waveform data to determine a comprehensive confidence coefficient; and counting the number of candidate targets of which the comprehensive confidence exceeds a threshold value in a specific time window, and determining a current rockfall risk level. Based on the method, video detection and vibration detection are combined, and the actual application effect is better.

A system and method for smart video detection is provided herein. The system includes an onboard processing unit having a smart video detection and extraction system. The smart video detection and extraction system recognizes and detects a target of interest within an image and / or video of a target object. The smart video detection and extraction system tags an image of interest within the target of interest to generate one or more flagged images and / or video. The onboard processing unit that sends the flagged images and / or video to a ground segment via a downlink.

The invention discloses an expressway traffic event video detection method, which relates to the technical field of video analysis, and comprises the following steps: acquiring a plurality of types of traffic event samples of an expressway, performing data calibration on an initial expressway traffic event training original data set according to multi-modal preprocessing, and obtaining an expressway traffic event training original data set; generating a highway traffic event multi-mode original training data set; performing spatio-temporal joint feature extraction on the multi-modal original training data set of the highway traffic events to obtain multi-modal spatio-temporal joint feature data, establishing a highway traffic event detection classification model, and evaluating probability distribution vectors of abnormal scores and event categories of the highway traffic events; and verifying the occurrence probability of the traffic event type under the given video stream and radar point cloud by using the probability distribution vector based on the abnormal score of the highway traffic event and the event type. The method has the advantages that the detection speed of the highway traffic incident is increased, and the detection accuracy of the traffic incident is improved.

The invention discloses a face video authentic identification method and system based on time sequence forgery clue analysis and a storage medium. The method comprises the following steps: acquiring a to-be-detected face video; and inputting a to-be-detected face video into the face counterfeit video detection model to obtain a face authentic identification result of each frame of image in the to-be-detected face video. Wherein the clues of short-term instantaneous anomaly, long-term gradual inconsistency and long-term accumulated distortion in a forged video are finely analyzed through associating an adjacent frame module, a future guidance frame module and a historical frame review module, so that the extraction capability of space general forged features is enhanced, and the accuracy of face authentic identification is improved.

The invention belongs to the technical field of vehicle overrun detection, and discloses a vehicle determination method and system based on video detection and distributed optical fibers, and the method comprises the steps: obtaining target information recognized by a camera, building a first database, obtaining target information recognized by the distributed optical fibers, and building a second database; pixel coordinates in the first database are converted into relative position coordinates, the similarity between the relative positions in the first database and the relative positions in the second database is judged within the same detection time, and if the similarity exceeds a certain threshold value, it is considered that data of the two vehicle tracks are from the same target vehicle; and fusing the monitoring data in the first database and the second database of the target vehicle to obtain an overrun detection database. According to the method, the target detection function of deep learning and the advantages of distributed optical fibers are combined, the vehicle determination method based on multi-device fusion is realized, and accurate determination of the vehicle is realized.

The invention relates to the technical field of machine vision detection, in particular to a video detection system for conveyor belt deviation and foreign matter accumulation, which comprises a video acquisition module, a self-adaptive illumination compensation module, a multi-scale feature fusion module, a space-time consistency analysis module and a decision output module. Wherein the video acquisition module is used for acquiring an original video stream of an operation area of a conveying belt; the self-adaptive illumination compensation module is used for generating an illumination equalization image through dynamic background separation and self-adaptive gamma correction; the multi-scale feature fusion module is used for extracting and generating a plurality of feature maps; the space-time consistency analysis module is used for outputting a deviation azimuth angle, a foreign matter coordinate and a confidence coefficient; and the decision output module is used for generating a deviation grade, a foreign matter type and an alarm instruction according to a preset threshold value. According to the invention, high-precision identification and classification judgment of the deviation of the conveying belt and the accumulation state of the foreign matters are realized through cooperation of multiple modules, and the intelligent level and the practical response capability of the detection system are improved.

A video detection model training method and apparatus, a video detection method and apparatus, and a device, relating to the technical field of artificial intelligence. The video detection model training method comprises: acquiring a video frame sequence, the video frame sequence comprising a plurality of video frames that are related to a target object and sorted in a chronological order (501); for m video frames in the video frame sequence, perturbing a key point of the target object in the m video frames to obtain a video frame sequence sample, there being an anomaly in the change of a feature associated with the key point of the target object in the plurality of video frames comprised in the video frame sequence sample (502); and training a video detection model on the basis of the video frame sequence sample (503). The present application perturbs key points to simulate phenomena "random jitter and drift of key points" present in fake videos so as to obtain samples, such that video detection models trained by means of the samples have a capability to detect "random jitter and drift of key points", thereby improving the detection accuracy of the video detection models.

A driving assistance device includes a line-of-sight information calculating unit to calculate line-of-sight information of a subject from a vehicle interior video, a target object information calculating unit to detect a target object present around the vehicle from a vehicle exterior video and extract target object information, a visibility calculating unit to calculate a visibility indicating a degree of visual recognition of the detected target object by the subject from the calculated line-of-sight information and the extracted target object information, a collision possibility calculating unit to calculate a collision possibility indicating a degree of collision possibility between the detected target object and the vehicle from the extracted target object information, and an effective collision possibility calculating unit to calculate an effective collision possibility from the calculated visibility and the calculated collision possibility.

The invention discloses a counterfeit video detection method and system based on selective multi-scale space-time fusion module capture, and is suitable for the field of video counterfeit detection. In view of the problem of long-time memory loss of a forged video and the problem that the difference between adjacent frames is difficult to perceive due to the fact that the forged video is always subjected to smooth frame-by-frame processing in practical application, local time information of different scales and difference information between two adjacent frames are respectively captured by constructing a double-branch time network; and therefore, the capturing capability of the model on subtle differences in forged video samples is enhanced. In combination with a space inconsistency module and an information supplement module architecture, the method realizes video space-time multi-scale fine-grained modeling, and realizes space-time information interaction. According to the method, the problem that in the prior art, only a single time feature or a single space feature is depended on is solved, the defect that the generalization ability of a model is insufficient is overcome, and the accuracy and robustness of forged video detection are improved.

The invention discloses a false news video detection method based on a retrieval enhancement and prototype alignment technology, which comprises the following steps of: processing a target video, constructing multi-modal information of the target video, and integrating the multi-modal information under a large language model to generate a unified text center query; performing video retrieval to obtain real and false video samples related to target video semantics; then respectively constructing prototype representations of real and false categories by using a graph attention network through a double prototype alignment mechanism, and generating a final operation perception representation through prototype alignment learning; and integrating the final operation perception representation with the existing model to complete the detection of the target video content. The subtle difference between the real news video and the false news video can be effectively identified, the detection of the video content is enhanced, the representation learning is realized, and the accuracy and robustness of the detection of the false news video are improved.