222 results about "Hash coding" patented technology

A high-quality digital human avatar is quickly reconstructed from a monocular video and is rendered in real time, the method is widely applied to various fields, and an existing method based on a neural radiation field is high in training and rendering cost. According to the dynamic human body three-dimensional modeling method based on the three-dimensional Gaussian sputtering technology, three-dimensional Gaussian can be expanded to a dynamic human scene, and rapid training and real-time rendering can be achieved. Specifically, the three-dimensional Gaussian of the human body is initialized in a standard space, and the three-dimensional Gaussian is converted from the standard space to a posture space by utilizing linear hybrid skin (LBS) according to posture parameters, so that fine details of the human body are learned at very low calculation cost; a posture parameter refining module and an LBS weight optimization module are designed to improve the problem of inaccurate detection of posture parameters in a real data set, and multi-resolution hash coding is introduced to accelerate the model training speed.

The invention discloses a quick retrieval method for e-commerce information, and relates to the technical field of e-commerce information.The method comprises the steps that features are extracted, multi-modal feature learning is conducted through image-text semantic matching and high-order semantic fusion, and commodity retrieval efficiency and accuracy are optimized through hash coding; analyzing and structuring user query through natural language processing and a deep semantic reasoning technology, and extracting a user intention in combination with numerical reasoning and logic processing; commodity matching is accelerated through Hash coding and approximate nearest neighbor retrieval, a retrieval result is optimized in combination with image-text semantic fusion and multi-factor sorting, and recommendation accuracy and personalization are continuously improved through a user feedback mechanism; the interpretability of a result is enhanced through reasoning chain echoing, and commodity recommendation is dynamically optimized by utilizing user feedback and reinforcement learning. According to the technical method, a more efficient, accurate and personalized commodity retrieval scheme is provided for the e-commerce platform, and the technical method has a wide application prospect.

The invention discloses a high-precision three-dimensional reconstruction method based on sparse depth, and the method comprises the steps: inputting an RGB-D image sequence of sparse depth, effectively filling the missing depth information in each frame of sparse depth image through a depth completion network trained on a public data set, and reasoning the semantic information of each frame through a pre-trained visual large model Dinov2; for each frame of RGB image, a combined sampling method based on feature point sampling and uniform sampling is adopted to extract sampling points of light, and spatial three-dimensional points are generated through importance sampling and random sampling strategies; performing position coding on the three-dimensional space sampling points to code low-frequency features, processing high-frequency features by using Hash coding, fusing the low-frequency features and the high-frequency features, inputting the fused high-frequency features and the fused high-frequency features into a neural network, and predicting a color value, an SDF value and a semantic value; and rendering and optimizing the pose of the current frame according to the current scene representation information, comparing a predicted value with the real color, depth and semantic information of the current frame by taking a global key frame as a sliding window, and optimizing implicit scene representation, thereby realizing reconstruction of a high-fidelity three-dimensional model from the sparse depth RGB-D image sequence. The method effectively solves the problem that a traditional three-dimensional reconstruction technology is insufficient in reconstruction precision under the sparse depth data condition, has a wide application prospect, and is particularly suitable for the three-dimensional reconstruction related field based on RGB-D data.

The invention discloses an image recognition method based on a deep neural network optimization algorithm, and relates to the technical field of image recognition, and the method comprises the steps: carrying out the preprocessing of an image data set, such as adaptive normalization and Hash coding label; coding a network structure according to a graph embedding technology, and mixing and initializing a weight threshold value; algorithm optimization features are fused, particle swarm initial determination is carried out, and then channel characterization fine tuning is carried out; using the training set to improve back propagation and adaptive moment estimation algorithm training, fusing a loss function and dynamically adjusting a learning rate; calculating evaluation and optimization of multiple indexes in the verification set, such as a genetic algorithm search structure and a Bayesian modulation learning rate; and inputting a to-be-recognized image into the optimized network, and obtaining a result by using an integrated recognition method. According to the invention, by optimizing the links of image preprocessing, neural network and recognition process, the image quality and data processing efficiency are improved, the features are accurately determined, the network training is accelerated, the network is comprehensively evaluated and optimized, the reliability of the recognition result is improved, and the image recognition precision and efficiency are improved.

The invention relates to a video frame data segmentation right confirmation and dynamic transaction method and system based on a block chain. The method comprises the following steps: segmenting video stream data according to a preset frame rate to generate independent frame images, and carrying out anti-collision hash calculation on the independent frame images to generate a unique identifier; generating a comprehensive value score through a multi-modal analysis model in combination with the independent frame image and the corresponding script text data; generating NFT assets anchored to the block chain according to the comprehensive value score and the unique identifier; and dynamically adjusting the income distribution proportion of the NFT assets through the smart contract based on the user film watching data. According to the method, uniqueness right confirmation of frame data is realized through precise segmentation and anti-collision hash coding, value evaluation is performed by using a multi-modal analysis model, and capitalization and on-chain right confirmation are completed by means of a block chain technology. In addition, the income distribution proportion is dynamically adjusted in combination with the smart contract, efficient transaction and right distribution of the video frame data are realized, and the commercial value and transaction efficiency of the video data are improved.

The invention provides a software demand defect detection method based on a large language model and a satisfiability problem solver, and the method comprises the steps: constructing a domain knowledge base, guiding the large language model to extract key elements in a demand text through combining with a cue word project, and recognizing imperfect and ambiguous defects; and meanwhile, the demand is converted into a logic expression, and logic inconsistency is detected by using an SAT solver. According to the method, incremental updating of the knowledge base is supported, efficient retrieval and comparison are achieved through vectorization and Hash coding technologies, experiments show that the method is remarkably superior to a traditional technology in demand detection in the fields of finance and the like, defects can be accurately positioned, correction suggestions can be provided, and the efficiency and accuracy of demand analysis are effectively improved.

The embodiment of the invention relates to the technical field of data storage, and discloses a cross-K8S cluster configuration change storage method and device based on a block chain, and the method comprises the steps: obtaining a transaction request message passing the compliance verification of an intelligent contract engine, and the transaction request message comprises a configuration change request; obtaining log information of the configuration change request in the transaction request message executed by the target K8S cluster, wherein the log information further comprises configuration change cluster difference information; the log information is stored in a fragmented mode based on the IPFS network, and content hash codes of the log information are obtained; and taking the transaction request message and the content hash code as complete transaction information, and storing the transaction request message and the content hash code in a transaction pool. Decentralized storage is achieved based on an IPFS network fragmentation storage mode, the problem that a centralized database or a log system is maliciously modified or historical records are deleted easily in centralized storage is solved, and auditing integrity and data storage safety are guaranteed.

The invention discloses an Internet of Things data storage method and device based on a block chain and a medium, and relates to the technical field of data storage, and the method comprises the steps: collecting an Internet of Things terminal data set, carrying out the noise disturbance and differential aggregation, and generating privacy protection data; performing hash coding on the privacy protection data, and outputting a tamper-proof to-be-stored data packet; performing security level encryption on the tamper-proof to-be-stored data packet by using a homomorphic encryption algorithm to form an anti-attack ciphertext vector; and carrying out data fragmentation and storage node mapping on the attack-resisting ciphertext vector to obtain a fragmentation storage index. According to the invention, through a dual privacy protection mechanism of homomorphic encryption and differential privacy, the privacy protection strength in the data storage process is improved. And meanwhile, safe and efficient multi-chain collaborative storage in a high-concurrency scene is realized through the constructed hierarchical cross-chain collaborative model.

The invention discloses an internet consumption analysis method based on data analysis, and relates to the technical field of data analysis, and the method comprises the steps: a plurality of participation platforms generate local cause-effect sub-graphs according to user behavior data, convert the local cause-effect sub-graphs into Hash codes, upload the Hash codes to a central server, and aggregate the Hash codes to generate a global cause-effect skeleton graph; injecting a predefined sequential logic rule based on the global causal skeleton graph, dynamically adjusting a causal edge weight according to a user real-time behavior event, and incrementally updating nodes and edges in the dynamic knowledge graph; pre-training the policy network parameters by using each participation platform, and uploading the pre-trained policy network parameters to a central server for aggregation; and issuing the aggregated policy network parameters to each participation platform, executing an interpretable policy according to the user real-time behavior event, and returning user feedback behavior data to each participation platform. According to the method, collaborative optimization of privacy security, real-time response and causal interpretability is realized through federal causal discovery and a dynamic knowledge graph evolution architecture.

The invention relates to an asteroid global terrain three-dimensional reconstruction method for illumination variation, and the method comprises the following steps: inputting an asteroid optical image, estimating the pose of a camera, generating a sparse point cloud, and initializing a Gaussian radiation field; using the image segmentation large model to output a segmentation mask of the asteroid optical image, and pruning the Gaussian radiation field; identifying a shadow region on the surface of the asteroid by combining a # imgabs0 # mixed mask output by a Gaussian radiation field; the camera position, the spherical harmonic coefficient and the Hash code are combined with the angle of the incident light to be input into a neural network, and color values of Gaussian primitives of illumination and shadow areas are regressed; depth normal consistency loss and rotation variance and scale constraint loss are added, optimization of a Gaussian radiation field is jointly supervised in combination with luminosity loss, a Gaussian opacity field is calculated, and a three-dimensional model of the asteroid global terrain is extracted. According to the method, the shadow region on the asteroid can be identified, and the three-dimensional reconstruction precision of the asteroid with surface illumination variation is improved.

The invention belongs to the field of artificial intelligence, and relates to a material retrieval method which comprises the following steps: acquiring a material sample set, and performing multi-modal feature extraction on each material picture of the material sample set to obtain text features and image features; splicing the text feature and the image feature of each material picture to obtain a plurality of multi-modal fusion features; performing Hash coding on the multi-modal fusion features to obtain material Hash codes of all the multi-modal fusion features, and constructing a retrieval database based on all the material Hash codes; obtaining a multi-modal fusion feature of the to-be-detected image to obtain a query vector, and calculating a hash value of the query vector to obtain a query hash value; and matching a material hash code most similar to the query hash value from a retrieval database to obtain a target hash code, and taking a material corresponding to the target hash code as a retrieval result. The invention further provides a material retrieval device, computer equipment and a storage medium. According to the method, quick retrieval of the material library is realized, and the accuracy of material retrieval is improved.

The invention discloses a dynamic airspace gridding management method and system for low-altitude economy, and belongs to the technical field of unmanned aerial vehicle traffic management, and the method comprises the steps: collecting airspace state data in real time through a multi-source sensing device, and constructing a four-dimensional space-time grid model; generating a four-dimensional space-time grid with a block chain hash code by fusing meteorological data, an airspace control rule and a real-time flight demand; receiving a space-time grid use request submitted by the aircraft through the smart contract, and calculating an optimal grid allocation scheme based on a deep reinforcement learning model; and the edge computing node executes local track prediction, issues a navigation instruction to the aircraft through the distributed account book, monitors a grid occupation state in real time, and triggers a dynamic grid recombination mechanism when sudden conflicts are detected. According to the method, the rigid constraint of static airspace division can be broken through, the cooperative conflict of multiple aircrafts is eliminated, and the marketization configuration of airspace resources is realized.

The invention discloses a conference content intelligent generation processing method and system based on a multi-modal large model, and the method comprises the steps: collecting the original data of a conference, and completing the standardization preprocessing; inputting a multi-modal large model, extracting multi-modal features and carrying out semantic alignment; executing cross-modal hash coding, generating binary codes and establishing an index database; performing hash retrieval on the related fragments, and constructing a conference content directed graph; based on a conference content directed graph structure, searching an optimized path by adopting a Monte Carlo tree; and generating structured conference content, and outputting a summary, an abstract and an action item. According to the method, efficient extraction, accurate retrieval and structured intelligent generation of the conference content are realized by fusing a multi-modal large model, cross-modal Hash coding and Monte Carlo tree search.

The invention discloses a neural radiation field rendering method based on dynamic hash coding, and the method comprises the steps: employing the feature sequence data as the input, calculating the density value and color value of each sampling point through the forward propagation of a neural network, carrying out the volume rendering integral operation according to the ray tracing principle in the direction of a ray, and obtaining the feature sequence data; judging a final color output result of the current pixel point; according to an error value between the color output result and a real image, updating a network parameter weight through a back propagation algorithm, and if the error value is greater than a convergence threshold, continuing to iterate the training process to adjust a feature coding strategy to obtain an optimized neural radiation field model parameter; and after the rendering performance configuration parameters are obtained, optimizing a storage allocation strategy of feature data through a memory pool management mechanism, and if the current memory occupancy rate exceeds a safety threshold, starting a data compression algorithm to reduce the storage space requirement, and obtaining a real-time rendering output result. According to the invention, high-quality real-time rendering of the dynamic scene is realized.

The invention provides a deep hash image retrieval method based on hierarchical multi-scale feature fusion. The method comprises the following steps: firstly, adopting a Nested Hierarchical Transform as a backbone feature extraction network, and synchronously capturing global context information and local detail features of an image through a nested local self-attention mechanism and a hierarchical feature aggregation module; then, constructing a multi-scale feature fusion module, extracting visual features of different receptive fields in parallel by using a multi-branch unequal-ratio expansion rate dilated convolution kernel, and adaptively weighting important features in combination with attention to generate discriminative multi-scale fusion features; and finally, designing a mixed loss function, optimizing intra-class feature compactness through center similarity loss, and realizing efficient and semantic similarity keeping Hash code generation in combination with a quantization loss constraint Hash code discretization process. According to the method, full expression of image features is realized through a collaborative mechanism of Transform and multi-scale feature fusion, and the precision and efficiency of large-scale image retrieval are improved through an end-to-end deep hash learning framework.

The invention discloses a large-scale embedded index data desensitization method suitable for a forestry database, and belongs to the technical field of data desensitization, and the method comprises the steps: obtaining a forestry data description document D1, splitting the document D1 into sentences, and processing the sentences into embedded vectors; a user customizes a query sentence containing the desensitization word, and a query vector is generated; performing hash coding on the embedded vector and the query vector to obtain a corresponding hash function cluster; searching data to be desensitized through cosine similarity calculation; and inputting to-be-desensitized data and the selected desensitization method into the large language model to generate desensitized data, and replacing the corresponding sentences in the D1 to obtain a desensitized forestry data description document D2. According to the method, the input data is converted into the high-dimensional vector through the embedding layer, so that the semantic features of the data can be effectively expressed in the vector space, the retrieval speed and accuracy of large-scale data are improved by means of the locality sensitive hashing technology, the calculation overhead is reduced, and a technical guarantee is provided for real-time desensitization processing of a large-scale data set.

The invention relates to the technical field of image processing and information retrieval, and discloses an image feature extraction and appearance trademark retrieval method based on deep learning, which comprises a feature extraction unit, a cross-modal fusion module, a retrieval strategy module and a dynamic weight distribution unit. The feature extraction unit extracts multi-level semantic information through a multi-branch convolutional network and multi-scale pyramid pooling; the cross-modal fusion module integrates vision, text and color information by using space attention, channel attention and cross attention mechanisms; the retrieval strategy module adopts Hash coding and global self-attention to realize rapid screening and fine matching; the dynamic weight allocation unit suppresses background interference through learnable parameters. The trademark retrieval precision and efficiency can be effectively improved, and the method is suitable for a multi-modal data processing scene.

The invention discloses a wafer defect real-time detection method and system based on multi-scale feature fusion, and belongs to the field of wafer defect detection, and the method comprises the steps: carrying out the preliminary coding of each scale feature through employing a convolutional neural network for a generated feature set, and obtaining a feature vector group containing the local and global information of a defect; if the saliency of the high-frequency component in the feature vector group exceeds a preset threshold value, performing weighted enhancement on the high-frequency feature through an attention mechanism to obtain an enhanced feature set; according to the enhanced feature set, constructing an index structure based on Hash coding, and generating a defect feature index table capable of being quickly retrieved by mapping feature vectors to a low-dimensional space; updating the index table according to the matching result, combining the newly detected defect features with the time sequence prediction result, and generating an expanded defect feature index table; and for the expanded index table, an incremental clustering algorithm is adopted to classify defect features, and defect type distribution updated in real time is obtained.

Disclosed is phishing classifier that classifies a URL and content page accessed via the URL as phishing or not is disclosed, with URL feature hasher that parses and hashes the URL to produce feature hashes, and headless browser to access and internally render a content page at the URL, extract HTML tokens, and capture an image of the rendering. Also disclosed are an HTML encoder, trained on HTML tokens extracted from pages at URLs, encoded, then decoded to reproduce images captured from rendering, that produces an HTML encoding of the tokens extracted, and an image embedder, pretrained on images, that produces an image embedding of the image captured. Further, phishing classifier layers, trained on the feature hashes, the HTML encoding, and the image embedding, process the URL feature hashes, HTML encoding and image embeddings to produce a likelihood score that the URL and the page accessed presents a phishing risk.

The invention relates to the technical field of inventory order auditing management, in particular to a big data-based non-correspondence auditing system and method.The big data-based non-correspondence auditing system comprises a data preprocessing unit, a data preliminary matching unit and a non-correspondence auditing unit. The data preliminary matching unit receives and processes data by using a message queue, partitions, de-duplicates and sorts the data, puts the data into a to-be-queued queue after preliminary matching by using a hash table structure, the non-corresponding auditing unit extracts features from historical data, establishes a dynamic threshold model through kernel density estimation, weights and fuses a threshold according to a data traffic load, and finally obtains a data stream; the deep learning model processes multi-dimensional features to output a normal delay probability to discriminate fuzzy data, and finally, a postponing area is set to review the fuzzy data according to a dynamic period to determine a final type, so that efficient data processing and accurate auditing are realized, data consistency is greatly improved, and enterprise operation decision scientificity is powerfully guaranteed.

The invention provides an AI-based photovoltaic module construction quality tracing system and method, and the system comprises an image collection module which obtains visible light, infrared and depth information, and forms original image data; the preprocessing module is used for carrying out geometric distortion correction, multispectral registration and noise suppression and outputting standardized image data; the feature extraction module is used for performing multi-scale feature fusion on the standardized image data to generate a multi-dimensional feature vector; the defect detection module is used for analyzing the multi-dimensional feature vector, detecting the defect type of the photovoltaic module and outputting a defect position coordinate; the construction parameter matching module is used for retrieving a construction process database according to the defect position coordinates and outputting a construction quality evaluation result; and the quality tracing module is used for performing hash coding, generating a tracing code and storing the tracing code to the block chain. According to the invention, the AI intelligent quality inspection robot is introduced and multispectral imaging is combined, so that real-time quality inspection of the photovoltaic module in the construction process is realized.

The invention discloses a depth image retrieval method and system based on metric learning and a storage medium, and relates to the technical field of image retrieval, a query image is input into an image hash coding model for image retrieval, and a similar image list is obtained; the training process of the image Hash coding model is as follows: constructing an image training set and a Hash center, performing feature extraction on batch images, and projecting the batch images to an embedding space to obtain feature vectors; projecting the Hash center to an embedding space to obtain a proxy; in the embedding space, the similarity between the feature vectors and the agents and between the feature vectors is optimized based on metric learning, so that the feature vectors converge to the same kind of agents, and the agents are adjusted to the same kind of feature vectors; reciprocity constraint is applied between the Hash codec and the Hash decoder, and measurement invariance constraint is applied to the Hash coding process, so that the Hash code is converged to a corresponding Hash center; according to the depth image retrieval method and system and the storage medium, the expression ability and retrieval precision of hash codes are improved.

The invention discloses an image retrieval method and a medium, and the method comprises the following steps: obtaining a to-be-retrieved image, and carrying out the feature extraction of the to-be-retrieved image, so as to obtain a depth feature map corresponding to the to-be-retrieved image; performing weighting processing on each channel of the depth feature map to generate a weighted feature map; extracting local features and global features of the weighted feature map, and performing feature fusion based on the local features and the global features to generate fused features; calculating a to-be-retrieved hash code corresponding to the fusion feature based on a hash-guided measurement loss method; and calculating the similarity between the to-be-retrieved image and the database image based on the to-be-retrieved hash code and the database image hash code, and determining a target image according to the similarity. Loss of important semantic information in the binaryzation process can be effectively prevented, and then the image retrieval precision is improved.

The invention discloses a semi-supervised low-rank cross-modal Hash retrieval method and a semi-supervised low-rank cross-modal Hash retrieval system for label noise. The method comprises the following steps: firstly, constructing an intra-modal and inter-modal contrast learning loss function, a structured semantic similarity loss function oriented to all samples, a cross-modal semantic correlation loss function and a label noise-oriented low-rank learning loss function, then constructing a total objective function, and respectively learning projection matrixes for a first modal and a second modal; based on the projection matrix, binary hash codes are generated for samples in the query sample set and the retrieval sample set; hamming distances from query samples to all samples in a retrieval sample set are calculated based on binary hash coding, samples corresponding to the first Z minimum Hamming distances are taken as retrieval results, and the problems that in the prior art, when training sample data are affected by label noise, effective supervision information is reduced, and the retrieval efficiency is improved are solved. Therefore, the technical problem that the trained cross-modal retrieval model cannot well mine identification features is solved, and the cross-modal retrieval precision oriented to the label noise influence is improved.

The invention relates to the technical field of digital health, in particular to an intelligent health data early warning method and system.The method comprises the steps that multi-channel signals such as electrocardio signals, pulse oximeter signals and continuous blood pressure signals which are synchronously filtered are injected into an erbium-doped annular cavity through light modulation to generate delay chaos, and sparse binary vectors are obtained through frequency domain hash coding; the vector predicts a future cardiovascular state through a spiking neural network and a reversible ordinary differential equation network; inputting the continuous prediction frame into a reversible tensor flow network, and obtaining a risk vector through anti-fact disturbance; splicing risks, states, codes and contexts to calculate risk energy, and outputting personalized intervention actions by means of a diffusion generation model; and synchronously updating model parameters and coupling coefficients according to the rewards. According to the invention, millisecond-level prospective early warning, low-power-consumption operation and interpretable intervention are realized, and the method is suitable for long-term wearable health monitoring.

The invention discloses an industrial smart energy multi-energy evaluation visualization system and method, and relates to the technical field of energy big data visualization crossover, and the method comprises the steps: collecting and preprocessing multi-source heterogeneous data, and generating space-time aligned high-dimensional energy data through a space-time hash coding method; based on the three-dimensional manifold coordinate model, converting equipment energy efficiency parameters into an equivalent energy mass field, and through an energy field reconstruction and geometric optimization method, generating a curved manifold space and identifying an optimal transmission path and a distortion strength cloud picture; and based on the distortion strength cloud picture, generating a thermodynamic color gradation through a thermodynamic color gradation mapping algorithm, rendering three-dimensional gradient field distribution of the equivalent energy mass field by adopting a ray tracing method, and generating a visual energy evaluation interface. According to the method, space and time hash codes are generated through a space-time hash coding method, the equipment-grid mapping relation is established in combination with GeoHash coding, and accurate space-time alignment of multi-source heterogeneous data is achieved.

The invention discloses a general model pre-training and self-adaptive optimization system and method for a remote sensing image in the technical field of remote sensing images. The method comprises the following steps: acquiring a structured data set containing multi-modal remote sensing core data and corresponding auxiliary geographic information; based on the structured data set, utilizing a two-channel feature extraction architecture to synchronously process visual spectral features and geographic spatial features, and performing model training in combination with innovative training tasks of geographic context contrast learning and multi-temporal mask reconstruction to construct a pre-training model with general characterization capability; and when the pre-training model is applied to a target domain, performing cross-domain feature distribution alignment by utilizing a dynamic gradient inversion layer, and selectively unfreezing part of model parameters by adopting a task perception unfreezing strategy to complete adaptive migration. According to the method, through a self-supervised pre-training architecture of geographic coordinate hash coding and multi-temporal mask reconstruction, the spatial-temporal feature extraction robustness is enhanced, and the problem of generalization of a small sample scene is solved at the same time.

The invention discloses a cross-modal retrieval method based on comparative learning and balanced Hash coding, which comprises the following steps: respectively extracting modal specific features and modal shared features from input multi-modal data, aligning the modal specific features of different modals through comparative learning, modal specific features and modal sharing features are optimized by using quantization loss based on optimal transmission; performing binarization operation on the modal specific feature and the modal shared feature to generate a modal specific hash code and a modal shared hash code; semantic clustering is carried out through a K-means algorithm based on the modal shared hash code to generate a semantic index; candidate samples are coarsely screened in the cross-modal retrieval stage through semantic indexes, then fine-grained comparison is conducted through generated modal specific hash codes and modal shared hash codes, and efficient cross-modal retrieval is achieved. According to the method, semantic alignment of different modes is realized by utilizing comparative learning, and the distribution characteristics of hash codes are optimized through the quantization loss based on optimal transmission, so that the retrieval performance is improved.

The invention belongs to the technical field related to volume video coding compression, and particularly relates to a multi-view video compression method and application thereof, and the method comprises the steps: initializing the information of each anchor point according to a video first frame; performing quantization and entropy coding on the current anchor point information of the partial anchor points by adopting a preset quantization offset parameter, and adding coding results to obtain Loss2; respectively adding corresponding quantitative offsets to current anchor point information of all anchor points, respectively and correspondingly inputting static and dynamic characteristics with offsets of each anchor point into mlp of neural Gaussian static and dynamic attributes to obtain static and dynamic attributes of each neural Gaussian of the anchor point, obtaining deformation attributes from the dynamic attributes, and obtaining deformation attributes from the deformation attributes; rendering based on the attribute information of each neural Gaussian to obtain a picture, and comparing the picture with a real picture to obtain Loss1; the mlp and all anchor point information are optimized through the two Loss, after optimization, the optimized anchor point information is quantized through Hash coding, entropy coding is conducted, and the coding result and the optimized mlp are compression results. The compression rate can be greatly improved.

The invention relates to a deep hash image retrieval method based on feature fusion and a dynamic converter, and relates to the technical field of artificial intelligence and computer vision, and the method comprises the steps of data preprocessing, feature fusion extraction, hash code generation and joint loss optimization. By combining the local feature extraction capability of CNN and the global modeling advantage of ViT, introducing dynamic convolution, a KAN module and a LoRA low-rank adaptation mechanism, and designing a joint optimized multi-branch loss function, the semantic consistency and retrieval performance of Hash codes are effectively improved, efficient and accurate large-scale image retrieval is realized, and the retrieval efficiency is improved. The problems that an existing image retrieval method is insufficient in feature expression, insufficient in retrieval precision and high in calculation overhead are solved.