16 results about "Task segmentation" patented technology

The application discloses a medical data processing method and product based on multi-stage transfer learning and multi-modal data collaborative fusion, relates to the technical field of medical data processing and artificial intelligence, and adopts MedicalNet medical special pre-training weights to initialize a classification model backbone network; a three-stage progressive fine-tuning framework is constructed on the basis, field adaptive coarse classification fine-tuning and target task fine classification are performed, and clinical structured data is introduced in the third stage; after high-dimensional image features are reduced in dimension and low-dimensional clinical features are increased in dimension through an adaptive multi-branch multi-level MLP architecture, mid-term deep fusion is performed, the application can effectively mine the complex relationships such as complementation, correlation and cooperation of multi-modal heterogeneous data, improve the lung disease classification precision and model generalization capability, and significantly inhibit the small sample overfitting phenomenon.

The application discloses a kind of blood vessels and lesion multi-task segmentation method based on ultra-wide-angle fundus image, belong to image processing field.The steps include as follows:S1.UWF fundus image dataset is obtained, and is assigned as training set and test set;S2.build multi-task semi-supervised learning network based on weight control mechanism, including encoder, decoder, cross-level non-local graph module and loss weight control mechanism, and training set image enters encoder and carries out feature extraction;S3.feature that encoder output is decoded and feature reconstruction by decoder part;S4.combined loss function and total loss are optimized to network by training, obtain the blood vessels and lesion multi-task segmentation model based on ultra-wide-angle fundus image;S5.data in test set are input into model, and segmentation result is obtained.The application can improve the ability of model to extract image detail features, realize fine and accurate multi-task segmentation.

The application discloses a layered accumulation floating-point matrix multiplication deterministic method, device and medium, relates to the field of high-performance computing and chip technology, and the method comprises the following steps: task segmentation is performed on a floating-point matrix multiplication calculation task according to a reduction dimension, and the floating-point matrix multiplication calculation task is distributed to multi-thread parallel processing under a plurality of computing clusters; intermediate results generated by thread operation are cached to shared memory of the computing cluster; a main thread of each computing cluster completes sequential accumulation of the intermediate results in the cluster according to a preset order to obtain a local accumulation result; and the local accumulation result is sequentially accumulated to global memory according to a predetermined order by controlling a global accumulation sequence number, so that the parallel calculation efficiency is ensured, the determinacy of the floating-point matrix multiplication operation result and the stability of the calculation precision are considered, and the application scene with high requirements on the reproducibility and precision consistency of the operation result, such as artificial intelligence reasoning and scientific calculation, can be adapted.

This invention discloses a method and system for assessing postoperative abdominal organ ischemia risk based on image analysis, belonging to the field of image analysis technology. The method includes acquiring enhanced abdominal CT images of the target patient and outputting them after standardization processing; employing an improved 3D U-Net++ architecture with multi-task collaborative learning to jointly segment the target patient's organs and target regions, integrating geometric priors and topological inference mechanisms; its key technical points are: using multi-task segmentation to focus RPPR calculation on the real ischemic area, avoiding average dilution of the signal across all organs, and topological correction to ensure that vascular features reflect the real anatomy; furthermore, through the joint analysis of low-perfusion area prediction masks and vascular VTIF, it reveals two ischemic subtypes: structural occlusion and functional hypoperfusion, promoting the individualization of clinical intervention strategies and defining the necessary vascular intervention or conservative treatment, making the overall solution both innovative and clinically applicable.

The present application relates to the technical field of data analysis, in particular to a self-adaptive task segmentation pipeline optimization method and system, which establishes a heat conduction parameter set of each computing node through the physical topology of a liquid-cooled computing cluster; collects the operation data of the computing node and the cooling system in real time, calculates the performance trajectory of each computing node according to the heat conduction parameter set and the operation data; when it is determined according to the performance trajectory that a computing node will trigger frequency reduction, a new task segmentation decision is generated after the unexecuted task flow is intervened cooperatively; the pipeline execution is regulated according to the new task segmentation decision; the execution efficiency of the pipeline task in the liquid-cooled computing cluster can be effectively improved.

This disclosure relates to branching operations for neural processor circuitry. A neural processor includes a neural engine for performing a convolution operation on input data corresponding to one or more tasks to generate output data. The neural processor circuitry also includes data processor circuitry coupled to the one or more neural engines. The data processor circuitry receives the output data from the neural engines and generates branching commands from the output data. The neural processor circuitry also includes a task manager coupled to the data processor circuitry. The task manager receives the branching command from the data processor circuitry. The task manager enqueues one of two or more segmented branches according to the received branching command. The two or more segmented branches are after a pre-branching task segmentation that includes a pre-branching task. The task manager transfers the task from the selected segmented branch among these segmented branches to the data processor circuitry for execution of the task.

This invention discloses an image recognition-based method for inspecting the construction quality of asphalt pavement, relating to the field of image recognition technology. The method includes: performing illumination normalization and texture feature modeling on a uniformly lit pavement orthophoto image to generate an illumination-normalized image and candidate anomaly response maps; inputting the illumination-normalized image into a multi-task segmentation network to perform multi-task segmentation inference, and combining it with the candidate anomaly response maps to perform differential operations to generate a reliable weighted defect probability map; performing temporal fusion and integral scoring on the reliable weighted defect probability map, and superimposing it onto the uniformly lit pavement orthophoto image to generate a quality inspection image report. This invention achieves efficient and automated inspection of asphalt pavement construction quality.

The application discloses a user interaction log-oriented internal and external noise filtering method, which firstly carries out screening and processing of an event log, and extracts useful column information; secondly, a general template in the form of a natural language text is automatically generated; then, the general log template and an operation line are combined, a Markov state transition graph of different operation types and target element combinations is constructed, and corresponding state transition probabilities are output; finally, internal and external noises are marked out through transition probabilities between nodes, and an event log is marked in detail by using a mean value and a total standard deviation variance adaptive threshold of a normal distribution, so that effective detection of internal and external noises in robot process automation is realized, and the accuracy of task segmentation, routine segmentation and process model generation is optimized.

The application discloses a medical data processing method and product based on multi-stage transfer learning and multi-modal data collaborative fusion, relates to the technical field of medical data processing and artificial intelligence, and adopts MedicalNet medical special pre-training weights to initialize a classification model backbone network; a three-stage progressive fine-tuning framework is constructed on the basis, field adaptive coarse classification fine-tuning and target task fine classification are performed, and clinical structured data is introduced in the third stage; after high-dimensional image features are reduced in dimension and low-dimensional clinical features are increased in dimension through an adaptive multi-branch multi-level MLP architecture, mid-term deep fusion is performed, the application can effectively mine the complex relationships such as complementation, correlation and cooperation of multi-modal heterogeneous data, improve the lung disease classification precision and model generalization capability, and significantly inhibit the small sample overfitting phenomenon.

The application discloses a kind of early warning method and device of children's eye disease and systemic genetic syndrome, it is related to ophthalmic examination equipment technical field.The method includes: collection ocular anterior segment image data set, utilize the multi-task segmentation model after training, obtain multiple congenital eye disease quantitative feature data, then based on the risk index calculator of pre-established index calculation is carried out, and the congenital eye disease risk index is obtained, classification processing is carried out, matches eye abnormal phenotype, and obtains systemic disease early warning report.The application can realize more than 10 kinds of congenital eye disease and rare but important lesion in one examination screening, and simultaneously establishes the intelligent early warning system of eye phenotype and systemic genetic syndrome.

This invention discloses an intelligent task allocation method for aircraft based on a classification-evaluation-matching decision tree. The method includes: constructing an aircraft system model and a task model; constructing and training a multi-level aircraft task allocation decision tree system; inputting the aircraft system model and task model into the trained decision tree system to generate a preliminary task allocation scheme; detecting time and fuel resource conflicts in the preliminary scheme and resolving conflicts using priority comparison, time shifting, and task segmentation strategies to form an executable task allocation scheme; and finally outputting the task allocation result after conflict resolution. This invention features high interpretability, comprehensive constraints, and high computational efficiency, effectively improving the completion rate and resource utilization efficiency of multi-aircraft collaborative tasks, and is suitable for intelligent task allocation in complex dynamic environments.

This invention discloses an AI-assisted medical image processing method and system, relating to the field of medical image processing. First, self-supervised learning is used to preprocess multimodal medical images, achieving denoising, registration, and standardization. Second, the processed images are input into a multi-task segmentation network based on a nested encoder-decoder architecture, simultaneously completing accurate segmentation of anatomical structures and lesion regions. Next, lesion features are extracted based on the segmentation results, and lesion localization is achieved through a 3D convolutional network and saliency map algorithm. A multimodal classification network is then used for benign / malignant differentiation and clinical staging assessment. Finally, the quantitative parameters of the lesions are automatically calculated, and all analysis results are integrated to generate a structured diagnostic report and provide multidimensional visualization. This invention addresses the problems of low computational efficiency, high system complexity, fragmented task correlation, and error accumulation in existing systems, achieving intelligent assistance throughout the entire process from image processing to clinical decision-making.

The application provides a method and device for constructing a double-task lung collapse evaluation model, which comprises the following steps: obtaining lung images with different lung collapse degrees, and marking the collapsed lung region and the chest cavity background region; extracting the collapsed lung region image and marking the collapsed region and the non-collapsed region label; training a lung volume task segmentation module and a lung color task segmentation module based on the marked image; constructing a fusion evaluation module for obtaining the lung collapse evaluation result based on the segmentation result of the lung volume task segmentation module and the lung color task segmentation module; and obtaining the double-task lung collapse evaluation model based on the fusion evaluation module and the trained lung volume task segmentation module and lung color task segmentation module. The application solves the problem in the prior art that the lung collapse evaluation mainly depends on the subjective experience of doctors during surgery, lacks artificial intelligence assistance, and results in large differences in scoring results, which easily leads to misjudgment.

The application discloses a kind of generative model heterogeneous collaborative inference method and system based on intelligent scheduling, it is related to distributed computing and artificial intelligence cross technical field, it is suitable for the cluster environment including multiple heterogeneous computing devices, method includes: the state data of each computing device is collected and normalized processing;The data after processing is input to the intelligent scheduling model trained unsupervised, and the non-uniform task segmentation ratio is output;Based on the ratio, the inference task of generative model is distributed, and the periodic synchronization strategy and fill-aggregation-cropping synchronization communication protocol are used to execute collaborative inference.The intelligent scheduling model is constructed based on graph neural network, the total delay of the system is predicted using the performance evaluation model based on Hockney model, and the delay is used as a loss function for model training, to achieve end-to-end optimization without labeled data.The application significantly improves the efficiency of collaborative inference and system resource utilization.

The present invention discloses a spatial grid data visualization system and method based on a three-dimensional model, which relates to the technical fields of computer graphics and real-time visualization. The method first analyzes the original grid data to identify the independent components of the model and their connection relationships; then constructs three independent representations: a topological skeleton layer, a parameterized bounding layer, and a geometric data layer; through quantization calculation and task segmentation, parallelly generates topological skeleton and parameterized bounding volume data, ensuring that the total generation time is less than a preset first-frame display time threshold; then submits for rendering to generate an initial three-dimensional image with complete component semantics; finally, while displaying the initial image, incrementally generates geometric data and performs progressive visual fusion with it; through hierarchical data organization and timing control, the present invention solves the problem of interface lag caused by the serial nature of data generation and display in traditional methods, and achieves a seamless transition from instant structure display to fine geometric rendering.

The application discloses a tongue image semantic reconstruction method fusing priori knowledge graph and image segmentation, relates to the technical field of intelligent tongue diagnosis, and comprises the following steps: organizing tongue state, tongue color, moss quality, tongue papilla, body fluid, crack, segmentation site and logical rules in the tongue diagnosis theory of traditional Chinese medicine to obtain a priori knowledge graph; acquiring an annotated tongue image, training a multi-task segmentation network in combination with the priori knowledge graph, and obtaining a knowledge-constrained multi-task segmentation network; inputting a pretreated tongue image into the knowledge-constrained multi-task segmentation network, performing tongue body segmentation, site segmentation and multi-feature extraction according to the segmentation site in the priori knowledge graph, and obtaining tongue body segmentation results, site segmentation results and feature distribution results. The application improves the stability of tongue body segmentation, site segmentation and multi-feature extraction by combining pixel constraint and semantic constraint through the knowledge-constrained multi-task segmentation network.