65 results about "Video quality" patented technology

Systems, methods, and software for operating an unmanned aerial vehicle (UAV). A method includes capturing a video stream using one or more camera sensors of the UAV. The method includes transmitting data representative of the video stream to a ground station communicably coupled to the UAV. The method includes determining one or more performance characteristics of a communications link between the UAV and the ground station. The method includes adjusting a resolution of the video stream according to the one or more performance characteristics of the communication link. Embodiments of the present technology provide continuity of video quality for viewing by UAV users at ground stations responsive to variations in video data transmission capacity during flight operations with minimal, or no, detrimental impact on user experience.

An image generation method and device, by forcing the alignment of the forward denoising distribution of the first graph model and the intermediate representation of the corresponding distribution of the second graph model, the first graph model retains the diversity advantage of the second graph model, while guiding the reverse denoising trajectory of the first graph model to accurately fit the reverse denoising trajectory of the first graph model, avoiding overexposure or blur problems, improving the generation quality, and finally realizing the dual optimization and balance of video quality and diversity.

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for selecting a video quality. One of the methods includes: determining, for a repeating time range of a video with a first video quality, a frequency with which one or more portions of the video were accessed, each portion for a corresponding one of one or more past instances of the repeating time range; selecting, for the repeating time range and using the frequency with which the portion of the video for the repeating time range was accessed, a second video quality from two or more video qualities that includes the first video quality; and storing, in memory, one or more additional portions of the video i) captured by a camera during future instances of the repeating time range ii) at the second video quality.

The application discloses an intelligent processing method, system and device for video merging and medium, and specifically comprises the following steps: dynamically dividing a to-be-processed video file into multiple variable segments; constructing a Merkle check tree according to the unique hash values of all the variable segments; obtaining first video data after segment transmission, extracting a visual feature vector of the first video data, identifying a subject object in a picture, performing three-level classification on a scene type and an action category, and generating second video data; performing frame position identification, key frame caching and time sequence alignment of multiple video segments on the second video data to form a merged video stream; eliminating joint traces of the merged video stream, monitoring video quality indexes in real time, dynamically adjusting quantization parameters of the merged video stream until the merged video stream reaches an output standard. The application improves the processing efficiency of video merging, greatly shortens the time consumption of large file merging, and meets the needs of users for efficient video merging.

The application provides a video intelligent synthesis method for multi-modal content conversion, comprising: acquiring video demand data, performing semantic analysis on the video demand data, generating a video intelligent synthesis data set, performing multi-path material retrieval based on the video intelligent synthesis data set, pruning in combination with a preset dynamic pruning strategy, generating a video synthesis material candidate set and a key frame demand set, generating an intelligent synthesis video segment based on the key frame demand set, performing quality screening on the intelligent synthesis video segment, generating a candidate video synthesis segment set, performing video intelligent synthesis based on beam search guidance according to the candidate video synthesis segment set and the video synthesis material candidate set, generating a candidate intelligent synthesis video set, performing quality evaluation on the candidate intelligent synthesis video set based on a preset video quality evaluation mechanism, and performing secondary synthesis according to the evaluation result to generate an optimal synthesis video, thereby improving the efficiency, accuracy and reliability of automatic video production.

This application provides a video quality assessment method, apparatus, electronic device, and storage medium. The method includes: acquiring a target video to be assessed; extracting visual characteristic parameters for each frame of the target video; determining human visual characteristic weights corresponding to each image region in each frame according to the visual characteristic parameters of each frame; determining distortion feature weights corresponding to each image region in each frame according to the visual characteristic parameters of each frame; weighting and fusing the distortion feature weights corresponding to each image region using the human visual characteristic weights corresponding to each image region in each frame to obtain a weighted distortion value corresponding to each image region in each frame; and determining a target video quality score corresponding to the target video based on the weighted distortion value corresponding to each image region in each frame and the temporal correlation between images in different frames of the target video.

The application discloses a dynamic bit rate allocation method and system considering parallel multi-channel video encoding, comprising the following steps: distributing the obtained multi-channel parallel power monitoring video stream to the corresponding video encoding channel based on the channel mapping relationship, and performing object domain type and encoder capability identification; determining the target object domain of the video stream, and determining the semantic importance level of the target object domain; combining the semantic importance level of the target object domain, the internal encoding complexity and the area characteristics to calculate the video stream complexity score; determining the code rate allocation strategy based on the video encoding channel attribute, solving the optimal code rate allocation scheme based on the code rate allocation strategy according to the video stream complexity score to obtain the target code rate; and matching the optimal key parameters of the video encoder according to the target code rate to compress and encode the video stream. The application can effectively balance the video quality of each channel, improve the code rate resource utilization rate and the encoding efficiency, and take into account the video content value.

Methods and systems, e.g., implemented by a client device, are provided for requesting a segment of a media content item using adaptive bitrate (ABR) streaming so as to improve the latter. To do so, a client device determines a first segment bitrate based on an available bandwidth. The client device then determines that a second segment bitrate is available, the second segment bitrate being lower than the first segment bitrate, and that a video quality of the segment of the media content item requested at the second segment bitrate is within a video quality variation range. Based on the determining that the video quality of the segment of the media content item requested at the second segment bitrate is within the video quality variation range, the client device requests the segment of the media content item at the second segment bitrate.

The application relates to the field of network live broadcast, and discloses a live video image coding method and device, equipment and a medium. The method comprises the following steps: determining whether to enable a fast decision mode according to the coding cost difference between a current image block in a current video frame of a live stream and the periphery image block which has been coded; when the fast decision mode is enabled, the luminance component and the chroma component of the current image block are used to judge whether the current image block belongs to a simplified coding skip block in sequence; when the current image block belongs to the skip block, the current image block is coded as corresponding skip information in the live stream; when the current image block does not belong to the skip block, a prediction coding mode of the current image block is determined, the current image block is coded into the live stream according to the best prediction coding mode determined by the determination; and the coded live stream is transmitted to a live broadcast room for display. The application can accelerate the coding speed by reducing unnecessary coding calculation, meanwhile, the video quality is maintained, and the coding efficiency and quality of the live video image are improved.

This invention discloses a method and system for evaluating the quality of immersive videos based on multimodal perception, relating to the field of video evaluation. The method includes: acquiring a multi-viewpoint texture and depth format video and extracting keyframes; extracting texture and depth features using ConvNeXt, processing texture features through a frequency-space texture enhancement module, and combining depth features with a cross-modal collaborative representation module to obtain structural texture coupling features; extracting semantic distortion features using a semantic distortion perception module; concatenating the coupling features and semantic distortion features and inputting them into a temporal modeling module to capture temporal features, then generating global perception features through a viewpoint fusion module; and finally outputting the final video quality score through a quality regression module. This invention achieves a more accurate and robust evaluation of immersive video quality by fusing multi-viewpoint texture and depth information and sequentially performing frequency-space texture enhancement, cross-modal collaborative representation, semantic distortion perception, temporal modeling, and viewpoint fusion.

The application discloses a kind of based on modular design's text video quality evaluation method and system, belong to computer vision technical field.The method includes: extracting time sequence, space and the multi-dimensional feature of graphic-text matching degree from original text video;Initial quality score of video is generated using multi-modal base model;Multi-dimensional feature and initial score are fused to generate prediction label;Through feature mapping block generation weight and bias coefficient, initial score is adaptively corrected to obtain final quality score;Using minimum batch learning strategy, combine prediction label dataset with real label dataset, to train quality evaluation model in a supervised learning manner.The application realizes the effective fusion of multi-dimensional quality characteristics through modular design, significantly reduces the dependence of the model on artificial annotation data, improves the consistency of evaluation results and human subjective perception, and provides an efficient and reliable solution for text video quality evaluation.

It achieves a sense of realism as if the voice is directly coming from the person on screen, at a low cost and without affecting the quality of the video. [Solution] The audiovisual device 1 reproduces and outputs audiovisual data as a video signal and multiple channel audio signals. Here, the audiovisual device 1 separates the audio band signal component from the center channel audio signal and outputs the separated audio signal along with the separated audio band signal component. The audio speaker 4 receives the audio band signal component output from the audiovisual device 1 and emits sound toward the screen of the monitor 2 connected to the audiovisual device 1. Here, the direction of sound emission from the audio speaker 4 is set so that the audio band signal component emitted from itself is reflected by the screen of the monitor 2 and reaches the listening point P.

This invention belongs to the field of video coding and image processing technology, specifically relating to a multi-level dual-stream pulse wavelet fusion panoramic video quality enhancement method, comprising the following steps: 1. Acquiring the original high-quality panoramic video image, performing video coding standard compression processing to generate a compressed low-quality panoramic video image; 2. Performing shallow feature extraction and mapping it to a high-dimensional feature space to obtain shallow features; 3. Performing feature splitting, inputting it into the parallel global and local stream branches of the pulse-driven biomimetic quality enhancement model SWFN for feature extraction; 4. Performing channel stitching, using fusion convolution for feature interaction and compensation, and outputting the fused features; 5. Performing residual connection and reconstruction, outputting a quality-enhanced panoramic video frame. This method can more accurately eliminate the block artifacts and quantization distortion caused by VVC coding compression, improving the reconstruction quality of ultra-high-definition panoramic video.

The application discloses a video encoding method, a video decoding method and device, equipment and a medium. When encoding a to-be-encoded video sequence of a portrait video type, original geometric parameters are mapped to a target geometric parameter set corresponding to a landscape video type, a video level is determined through the target geometric parameter set and a video level limit parameter, and the to-be-encoded video sequence is encoded with the original geometric parameter set. When decoding, the video level determined through the target geometric parameter set is used for decoding. This method of determining the video level through the target geometric parameter set and encoding / decoding can avoid the situation that the original geometric parameters of the portrait video type exceed the video level limit parameter, resulting in the situation that encoding / decoding cannot be performed. The application also avoids the loss of the quality of the original video, and realizes the encoding / decoding of the portrait video by using a low video level encoder / decoder without increasing the processing burden and hardware cost and without affecting the video quality.

The deep learning-based video multi-level adaptive compression cross-layer transmission optimization method and system of the application relates to the technical field of video transmission. By designing a compression level selection model based on deep learning, the optimal compression level is output based on the original video data and network environment parameters, and the corresponding compression strategy of the optimal compression level is obtained. A transmission strategy selection model is designed to generate a transmission strategy based on video quality parameters and network environment parameters. A global utility function is defined to calculate the global utility value when the compression strategy and the transmission strategy are assumed to be executed. An optimization objective function is designed, which inputs the compression strategy and the transmission strategy and outputs the optimized compression strategy and the transmission strategy. A condition judgment logic is set to select the final strategy based on the compression strategy and the transmission strategy before and after optimization, thereby achieving global optimization of video compression and transmission.

The application discloses a method for evaluating the quality of UHD images and videos based on graph convolution and space-time correlation, which comprises the following steps: dividing an input ultra-high-definition image according to a preset grid, and obtaining node features with uniform dimensions through linear mapping. Based on the spatial distance between the normalized central coordinates, the node features and a normalized adjacency matrix are input into a multi-layer residual graph convolution network for feature propagation to obtain node representation. The node representation is subjected to attention weighted pooling to obtain a standardized quality prediction value. The quality prediction value is subjected to learnable affine calibration, and is subjected to inverse standardization according to statistical parameters of quality scores of a training set to obtain a final image quality score. The method can adapt to the high-dimensional input characteristics of ultra-high-definition images, and can combine the time sequence correlation between video frames to realize a high-precision and high-time-efficiency method for evaluating the quality of ultra-high-definition images and videos without reference.

The application provides a kind of digital human enhancement rendering method and system based on low-resolution video, related to digital human technical field, in view of the problem of poor quality of low-resolution digital human video in the prior art, by obtaining the low-resolution digital human video to be processed, the low-resolution digital human video is composed of continuous image frames containing digital human form and motion information. A digital human rendering feature field interaction model is constructed to represent the dynamic influence relationship between different frame shape and motion characteristics, generate inter-frame rendering coordination rules, and specify the inter-frame feature linkage adjustment mode. According to this rule, the image frame is executed frame by frame associated rendering, and the enhanced image sequence with inter-frame feature coherence is generated, and the high-definition digital human video is outputted according to the original frame timing, which effectively improves the quality of low-resolution digital human video.

This invention relates to a method and medium for quality assessment of text-guided image-generated video. The method includes: firstly, extracting multi-source heterogeneous features from the original image, text guidance, and generated video to construct a RAG reference library containing millions of video clips. Then, using the InternVL3 multimodal model, unified feature encoding is performed on the image, text, and video to achieve multi-dimensional assessment of text-image-video semantic consistency, dynamic semantic preservation, and text guidance compliance. The system dynamically calculates quality scores from each data source, integrates multi-dimensional assessment indicators, and outputs a comprehensive quality score. This invention achieves comprehensive, objective, and dynamic assessment of text-guided image-generated video content, solving the problems of single assessment indicators and lack of dynamic semantic preservation assessment capabilities in existing technologies, and providing a reliable quality assessment tool for the optimization and application of image-generated video technology.

The application relates to the technical field of video diagnosis, and particularly discloses a video quality diagnosis system and method based on a multi-modal large model, which comprises a data acquisition module, a data set construction module, a model fine-tuning module, a preliminary diagnosis module and a deep diagnosis module, constructs a professional knowledge enhanced data set of a current video quality diagnosis process, fine-tunes a visual-linguistic base model in a field by using the professional knowledge enhanced data set, obtains a field video quality diagnosis model, inputs video frame sequences and equipment metadata in a to-be-diagnosed original video stream into the field video quality diagnosis model, and outputs a preliminary diagnosis result; deep root causes leading to fault phenomena in the preliminary diagnosis result are inferred to generate a structured comprehensive diagnosis report; and the application can improve the diagnosis efficiency of video quality.

The present application belongs to the field of textile intelligent manufacturing and dyeing and weaving process optimization, and particularly relates to a kind of textile dyeing weaving process intelligent design method and system based on Agent.The method constructs knowledge base containing mechanism model, process template, constraint and quality association information, forms feature input by fusing order and grey cloth, equipment capacity, historical batch and image / video quality data, generates candidate process scheme by using Agent multi-stage reasoning, and obtains executable process that can be directly issued by constraint checking and linkage repair through device window and timing sequence;At the same time, the quality backtracking evidence chain and version iteration mechanism are introduced, and the abnormal attribution, grey verification and rollback update are realized.Compared with the prior art, the present application can improve the process design efficiency, reduce the trial and error and quality fluctuation risk, and improve the adaptation ability to new materials and new equipment.

This invention relates to the field of cellular-connected drone video transmission technology, specifically disclosing a method for maximizing user experience quality in multi-drone aerial video transmission. This method is used in aerial video streaming systems that utilize multiple cellular-connected drones to capture video from different points of interest (PoI) regions and transmit the video to ground base stations (BS) and users, allowing ground users to share the drones' field of view. Based on this aerial video streaming system, this invention constructs an optimization problem. By jointly optimizing transmission scheduling, video playback rate, and drone trajectory design, it maximizes the minimum quality of experience (QoE) for all users, considering uplink interference and the trade-off between video quality and smoothness. Simulation results show that, compared to benchmark solutions, the proposed solution achieves a significant performance improvement and achieves a trade-off between video quality and smoothness.

Embodiments of the present disclosure disclose a video generation method, device, electronic equipment, computer readable medium and program product. A specific implementation of the method comprises: obtaining a video frame sequence, the video frame sequence being obtained by frame-level splicing of a time-sequentially continuous video segment sequence, the video segment sequence being generated based on a target audio and a target person image; determining the video frame sequence as a target video frame sequence, and performing the following processing steps: dividing the target video frame sequence into frame groups to obtain a frame group sequence; performing smoothing processing on the frame groups in the frame group sequence to obtain a new frame group sequence; and in response to determining that the new frame group sequence meets a preset requirement, generating a video corresponding to the new frame group sequence. This implementation is related to generative artificial intelligence and helps to improve the quality of the generated video.

The application provides an error-resistant video coding method and system based on channel importance perception redundancy allocation. The method comprises: generating a quantized feature tensor of a current frame by using a neural video codec; performing importance evaluation on each channel of the quantized feature tensor based on an intra-distribution region length criterion to calculate an importance index of each channel; adaptively determining the number of important channels that need to be protected by using a cumulative importance summation function according to the packet loss rate of the current channel; retaining important channels, eliminating unimportant channels, and reallocating the bit rate resources released by the elimination of unimportant channels to important channels to improve the anti-packet loss capability of important channels through multiple copy methods; and setting the lost channel elements to zero at the decoding end and reconstructing the video frame by using an error-resistant decoder. The application has a significant performance advantage compared with existing methods under an extreme packet loss scenario (packet loss rate exceeding 60%), and exhibits superior robustness and video quality maintenance capability.

The application discloses a kind of video generation object quantity control methods based on layout guide without training, the method includes: using the self-attention and cross-attention graph in the video pre-generation process, screening best attention head and constructing initial semantic layout to identify object quantity;According to the difference between current quantity and prompt word target quantity, by minimizing cost function, add and delete correction is carried out to layout graph at instance level;Using the corrected layout graph as guide signal, modulate cross-attention mechanism to regenerate video.The application uses the "first identification-then guide" paradigm without training, effectively solves the counting error problem caused by weak semantics and instance ambiguity of existing text-to-video model, while maintaining video quality and temporal coherence, significantly improves the compliance ability of generated video to quantity instruction.

The application relates to a video generation method and device, computer equipment and a storage medium. The method comprises the following steps: subject classification is performed on video materials, thereby obtaining a plurality of subject events; each material segment in each subject event is scored, thereby obtaining quantized material scores; a plurality of material segments with high material scores are screened out in combination with video scene elements corresponding to a narrative style of the video materials; and a final video product is generated, without manual editing, and with different subject events logically arranged, so as to reflect the logical connection between different material segments, select material segments with high material scores in different subject events to generate a video product, improve the video quality of the video product, and solve the problem that existing video editing tools cannot understand the internal connection between materials, resulting in a lack of logic and story in the video product.

With video quality estimation provided as an operating system service or cloud service, estimates of video quality can be collected unobtrusively and without feedback from video playback applications or viewers. For example, for a portion of reconstructed video content, an operating system service of a client computer system receives video data, estimates video quality of the portion of reconstructed video content using the video data, and sends results of the video quality estimation to an application executing on the client computer system. Or, as another example, for a portion of reconstructed video content, a cloud service of a server computer system receives video data, estimates video quality of the portion of reconstructed video content using the video data, generates encoder control values based at least in part on analysis of results of the video quality estimation, and sends the encoder control values to a streaming or conferencing service.

A method and an apparatus are disclosed for video coding for refining predicted signals of intra prediction based on deep learning. The video coding method and the apparatus generate a refined intra predicted signal to improve video coding efficiency and video quality. The video coding method and the apparatus adaptively input a prediction block according to intra prediction of a current block, block information related to the current block, and neighboring reconstructed reference samples into a deep learning network.