8results about How to "Valid reservation" patented technology

The embodiment of the application provides a method for converting a knowledge graph into a natural language large model training sample and related equipment, and belongs to the technical field of artificial intelligence. The method comprises the following steps: selecting a key anchor point from the knowledge graph based on a graph centrality measurement index; taking a key anchor point pair as a starting point and an ending point, extracting a symmetric element path connecting the two, and constructing a subgraph; generating multi-perspective paths for the same core entity through an anchor point switching mechanism; converting subgraph information into a (instruction, output) sample pair conforming to the instruction fine-tuning paradigm by using a structured information template, a path information template and a metadata enrichment template; and fine-tuning a large language model by using the generated corpus. By introducing the symmetric element path and the multi-dimensional instruction template, the embodiment of the application realizes semantic preservation and structured mapping of complex relationships in the knowledge graph, and effectively improves the knowledge understanding, reasoning and generalization ability of the large model in a specific field such as the power secondary equipment field.

The utility model discloses an inflator for inflating product belongs to inflating device technical field, and it includes the cylinder, the inflator and the piston structure. Its innovation lies in setting up the oil absorption cotton structure, and this oil absorption cotton can be installed in two positions: one is on the piston structure, and specifically can be set in the positioning groove of piston or the annular groove specially, is located the rubber ring inside, two is the bottom of the cylinder. The piston structure includes the support disc with the positioning groove and the annular groove and the rubber ring with the inverted L type section. The oil absorption cotton can be annular or circular, and can be provided with tail strip to facilitate replacement. The utility model prolongs the service life of inflator, improves the inflating cleanliness through the effective adsorption and slow -release lubricating oil of oil absorption cotton, reduces the friction wear and oil spill, is especially suitable for the product such as inflating toy.

This invention belongs to the field of food processing technology and discloses a method for producing anthocyanin-enriched antioxidant jelly from black chokeberry. By weight, the raw materials include: 6.5-19.5 parts black chokeberry, 3.5-10.5 parts persimmon, 7-21 parts erythritol, 0.4-1.2 parts carrageenan, 0.4-1.2 parts konjac gum, and the remainder being drinking water. The method employs a persimmon compound astringency removal process to eliminate the astringency of the black chokeberry itself; uses erythritol, a natural sugar substitute, to replace sucrose as a sweetener, which reduces calories and avoids masking the natural flavor of anthocyanins; and utilizes a compound gum to form the jelly, combining the anthocyanins of black chokeberry with the jelly; thus developing a low-sugar, healthy, functional food to meet consumers' demand for healthy snacks.

The application discloses a multi-domain modeling and semantic embedding enhanced lightweight small target detection method. The multi-domain modeling and semantic embedding enhanced lightweight small target detection method comprises the following steps: acquiring a to-be-detected image; inputting the to-be-detected image into a trained lightweight detection model, so as to determine a backbone feature map of the to-be-detected image through a plurality of high-frequency residual enhancement spatial depth conversion convolution units and a plurality of spatial-frequency context enhancement layer aggregation units; fusing the backbone feature map through a feature pyramid fusion method to obtain a fused feature map; determining a neck feature map based on the fused feature through a plurality of frequency-guided semantic injection units; and detecting the neck feature map through a plurality of detection heads to obtain a detection result of the to-be-detected image, thereby improving the detection precision of small targets.

This invention provides a protective mobile unmanned surface vessel (USV) hyperspectral system for shallow sea pollution observation, belonging to the field of marine environmental monitoring technology. It includes an USV hull, hyperspectral observation equipment, a water sampling and detection device, and a controller. By installing a cabin and a hinged door on the USV hull, and housing the height-adjustable hyperspectral observation equipment, the system effectively protects the equipment when observation is needed. The door opens to extend the equipment from the cabin, and retracts it when observation is not required, then closes the door. The water sampling and detection device allows for timely and rapid sampling of polluted water, analysis of the samples, and output of results. This invention integrates hyperspectral observation and water sampling and detection, effectively protecting the hyperspectral observation equipment while simultaneously enabling sampling of polluted water, thus improving the effectiveness of pollution evidence collection.

This invention discloses a method for detecting minute defects in cigarette boxes based on a CDM-YOLOv8 network, comprising: Step 1, acquiring an image of the cigarette box surface and constructing a defect detection dataset; Step 2, performing feature extraction enhancement on the input image, extracting defect-related features through multi-branch convolution, deformable convolution, and spatial attention mechanisms; Step 3, dynamically fusing features at different scales, and performing frequency domain banding processing on the fused features during the downsampling stage, employing a differentiated downsampling strategy to preserve defect edge and texture information; Step 4, outputting the category and location information of the defects based on the detection head. This invention solves the problems of easy loss of minute defects in cigarette boxes during downsampling and easy missed detection in complex backgrounds. By enhancing feature representation capabilities and preserving high-frequency defect information while ensuring computational efficiency, it improves the detection accuracy of minute defects in cigarette boxes and reduces the missed detection rate.

A garment color generation system and method for human body images belong to the field of garment auxiliary design. In view of the problem of garment automatic coloring, a human body image dataset is first constructed, including original human body images, local gray human body images and corresponding human body analysis images; then a locally controllable garment color generation system is designed, including a color feature extraction module and a color generation adversarial network. The color feature extraction module extracts the color matching scheme from the reference material and generates the color feature map, realizing the local controllability of color rendering. The color generation adversarial network includes a generator with a "double encoder-single decoder" structure and a discriminator, and the generator fuses human body features and color features to automatically color the garment area in the human body image. The color generation system adopts the generator and discriminator in an adversarial manner for training, and is matched with various auxiliary loss functions, realizing high-quality human body image generation.

This invention provides a method and system for suppressing seismic network noise in marine seismic data, relating to the field of seismic network noise suppression technology. The method includes acquiring marine seismic data and performing single-shot input processing to form initial single-shot seismic input data; locating network noise based on amplitude characteristics of the initial single-shot seismic input data to pinpoint the network noise; and suppressing the network noise to form suppressed single-shot seismic data. This method effectively suppresses network noise by applying median filtering and linear suppression processing to the seismic data.