12results about How to "Efficient Supplement" patented technology

The present application belongs to the technical field of computer vision three-dimensional reconstruction, and specifically relates to a semi-supervised single-view 3D object reconstruction method. A small amount of labeled samples are used to train a neural network, and then the trained neural network is used to generate pseudo labels for unlabeled samples and guide the training of the unlabeled samples. At the same time, the present application proposes that a discriminator scores the quality of the generated pseudo labels, and limits the bias of low-quality pseudo labels on model training. In addition, the present application proposes a prototype shape prior module based on an attention mechanism, which serves as a bridge between images and 3D shapes, reduces the difference between the two modalities, and provides shape prior for the neural network, ensuring that the reconstructed 3D shape is natural. Compared with the current mainstream method in the industry, the present application is more accurate in three-dimensional reconstruction of a single image under the condition that the amount of labeled data is very small, and the generated 3D shape is more realistic and natural.

The invention discloses a multi-coal-source coal blending data processing method based on a WSMOTE algorithm, and belongs to the technical field of coal processing data processing, and the method comprises the steps: data collection and classification: collecting multi-dimensional data of a raw coal floating and sinking test, a coal blending test and the like, carrying out the preprocessing, dividing the data into a missing data set and a complete data set, carrying out the interpolation optimization of a missing value, and carrying out the calculation of the complete data set. And carrying out adaptive processing according to data distribution types, or carrying out weighted combination normalization according to feature importance, analyzing the number and distribution characteristics of minority class samples, dynamically adjusting the neighbor sample size and the new sample synthesis amount, and completing data enhancement and standardization. According to the method, standardized parameters are dynamically adjusted to adapt to data distribution changes, abnormal values are accurately processed, the combined interpolation model is subjected to weight optimization, errors are reduced, and data integrity is improved. WSMOTE parameters are adaptively combined with data distribution to synthesize samples, data are balanced, and the over-fitting risk is reduced. In practical application, the quality of coal blending model training data can be improved, and optimization of a multi-coal-source coal blending scheme is assisted.

A multi-groove cooperative type end face mechanical sealing structure comprises a rotating ring and a static ring which are used for mechanical end face sealing, the rotating ring and the static ring are both provided with annular sealing end faces which are coaxially arranged, and the sealing end faces take the circle center of the sealing ring as the reference. An outer side sealing face, an annular groove and an inner side sealing face are sequentially arranged in the radial direction from the outer edge to the circle center. The outer side sealing face is provided with drainage grooves distributed in the circumferential direction, the drainage grooves penetrate through the whole outer side sealing face, the inlet ends of the drainage grooves are connected with the medium side, and the outlet ends of the drainage grooves are connected with the annular groove. When the rotating ring rotates, a medium enters the drainage groove from the inlet end of the drainage groove and then flows into the ring groove from the outlet end. Through the drainage groove, the pressure of a medium in the annular groove can be greatly improved, the medium can be effectively supplemented to the inner side sealing face, the continuity of formation of a liquid film on the inner side sealing face is ensured, and dry friction of the inner side sealing face and flash evaporation of the medium caused by too high temperature are avoided.

This invention discloses a fine-grained multimodal named entity recognition method for knowledge extraction in flight control experiments, belonging to the field of aviation test data processing and artificial intelligence fusion technology. The invention includes acquiring text and image data from flight control test documents; extracting text feature sequences and visual feature sequences; performing semantic enhancement based on a pre-built tag prototype memory, which contains multimodal prototype representations of entities of each category extracted through clustering algorithms; calculating cross-modal attention distribution and generating intramodal association matrices through mutual construction operations; performing feature fusion based on the intramodal association matrices; and performing sequence decoding on the multimodal joint features to output entity category labels. This invention solves the problem of existing methods lacking reliable guiding signals when text information is insufficient, overcomes the limitations of the traditional one-to-one alignment assumption, and can explicitly model the complex many-to-many semantic mapping relationship between text and visual objects in flight control test documents, achieving a deeper level of semantic fusion.

The invention discloses compact anti-wrinkle elastin peptide freeze-dried powder and a preparation method thereof, and belongs to the technical field of bioactive peptides. The freeze-dried powder is prepared by taking core tubes and / or throat tubes of bred animals as raw materials through cleaning and impurity removal, crushing and homogenizing, degreasing and impurity removal, composite enzymolysis, enzymolysis inactivation, centrifugal separation, gradient filtration, concentration, sterilization and vacuum freeze drying, and has remarkable elastinase inhibitory activity and tightening and anti-wrinkle effects. The livestock and poultry slaughtering byproducts which are wide in source and low in cost are selected as raw materials, high-value utilization of the livestock and poultry slaughtering byproducts is achieved, the optimized preparation technology solves the problems that in the prior art, enzymolysis efficiency is low, product purity is low, activity is insufficient, and stability is poor, the technology is simple, large-scale production is easy, and the method can be widely applied to the fields of cosmetics, food and health care products.

This invention relates to a method for preparing potassium-enriched soy sauce, comprising the following steps: adding food-grade potassium salts to a fermented and extracted soy sauce base liquid for fortification, so that the potassium content of the final product reaches 3500-4000 mg / 100 mL. (a) Raw material processing: soaking and steaming soybeans, roasting and crushing wheat, and mixing the two; (b) Koji making: inoculating the mixture with Aspergillus oryzae for cultivation to obtain koji; (c) Fermentation: mixing the koji with brine for dilute mash fermentation; (d) Extraction: performing solid-liquid separation on the fermented mash to obtain the soy sauce base liquid. This invention significantly increases the potassium content of soy sauce to approximately 3700 mg / 100 mL, classifying it as a high-potassium fortified food. For people with "low potassium constitution" who do not consume enough potassium in their daily diet, there is no need to deliberately increase the intake of vegetables and fruits; potassium can be conveniently and efficiently supplemented simply through daily cooking and seasoning.

The application discloses a compressed air energy storage system suitable for multiple working conditions and a running method thereof, comprising an isothermal compression system, a gas storage system, an expansion work system, a molten salt heat storage system and an air-flue gas heat exchange system; during a low electricity consumption valley, low-valley electricity is utilized to drive the isothermal compression system to compress air, and the compressed air is stored in an underground salt cavern; during an electricity consumption peak period, compressed air in the salt cavern is released through an electromagnetic valve, and is divided into two work routes: one route is provided with supplementary combustion, and the compressed air is expanded to work twice; the other route is not provided with supplementary combustion, and the compressed air is expanded to work once. The application stores compressed air by utilizing low-valley electricity, and releases energy to generate electricity during the electricity consumption peak period. The system is flexible and efficient, is an effective supplement to an existing energy storage system, and has high practical value.

This invention relates to the field of environmental perception and discloses a pure visual 3D target detection system and method based on a bird's-eye view. The pure visual 3D target detection system based on a bird's-eye view includes: a 2D image feature extraction module extracting features from the original image to obtain 2D image features; a 2D semantic information fusion module fusing the 2D semantic segmentation mask with the 2D image features to obtain 2D features with fused semantic information; a depth feature extraction module combining the 2D image features to obtain 2D features with depth information; a BEV feature projection module projecting the 2D features with depth information onto the BEV feature space to obtain the current frame BEV features; a temporal information fusion module splicing and fusing the aligned historical frame BEV features with the current frame BEV features in the channel dimension to obtain BEV features with fused temporal information; and a 3D target detection module outputting 3D target detection results.

The invention discloses a method, a system, equipment and a medium for recording an original voice stream in real time and converting the original voice stream into a playable voice file, and the method comprises the following steps: after a calling party and a called party establish a call, combining SDPs of the calling party and the called party into the same SDP capable of distinguishing the calling party and the called party, sending an SIP signaling to a naked stream recording server through an SIP protocol, after a recording request is received, a calling SDP and a called SDP in the SIP signaling are analyzed, a media format is obtained through media level description and fmtp in the SDP, and received voice media streams are classified and stored according to the media format; the media format is divided into constant packet media and non-constant packet media; for the non-constant packet media, a mute packet is supplemented to the voice media stream; respectively compressing and packaging the constant packet media and the non-constant packet media supplemented with the mute packets to obtain a playable audio file; and combining the playable audio files of the calling party and the called party into a dual-track audio file. According to the invention, the system performance consumption caused by transcoding and frame supplementation required by original recording is reduced, and the concurrency performance of stand-alone voice recording is improved.

The utility model discloses a sewage interception flow guide pipe which comprises a flow guide pipe and a filter element device inserted into the flow guide pipe, the flow guide pipe comprises a flow guide pipe shell, and the filter element device is inserted into the flow guide pipe shell, is supported and fixed at the top of the flow guide pipe shell and is used for guiding and filtering rainwater and intercepting foreign matters; the filter element device comprises a filter element shell and a plurality of layers of filter parts regularly arranged in the filter element shell. According to the sewage interception flow guide pipe disclosed by the utility model, rainwater is effectively intercepted and filtered through the filter element device inserted into the flow guide pipe. One part of the filtered rainwater is directly absorbed by root systems of vegetation, and the other part of the filtered rainwater can effectively supplement underground water. Meanwhile, the filter element device can be conveniently taken out from the flow guide pipe to be replaced, a plurality of filtering parts in the filter element device can be conveniently replaced and cleaned, and the use cost is greatly reduced.

The invention discloses a para-aramid solvent recovery control method which comprises the following steps: fully mixing an NMP (N-Methyl Pyrrolidone) replenishing solution with a polymerization mother solution to obtain a mixed solution; the polymerization mother liquor is an aqueous solution obtained after PPTA resin is polymerized and washed; carrying out refining treatment on the mixed solution to obtain an NMP (N-Methyl Pyrrolidone) raw material; mixing the NMP raw material with a CaCl2 solution with the concentration of 38-40% to obtain a polymerization solvent; and returning the polymerization solvent to the polymerization reaction of the PPTA resin for continuous use. According to the control method for recovery of the para-aramid solvent, effective supplement of NMP in the polymerization solvent and effective control of moisture of a solvent system can be ensured in continuous production.

The application discloses a subject corpus annotation method and system based on classifier multi-round iteration, and belongs to the field of natural language processing and machine learning. The method comprises the following steps: constructing a multi-level subject classification system, based on a Fasttext classifier and a seed set, performing multi-round iterative reasoning on a target subject, dynamically generating positive and negative samples in each round of iteration, and gradually improving the classification accuracy; using a large model to score the positive samples in multiple dimensions to screen high-quality positive samples, and extracting keywords to optimize the classification boundary; generating subject questions through a WebQA method and retrieving supplementary low-recall rate subject corpus; training an information filter to identify and filter out low-quality content such as advertisements and garbage; and finally realizing efficient and accurate subject annotation through the series process of the classifier and the filter. The application solves the problems of low efficiency, large resource consumption and subject understanding deviation of traditional annotation, and is suitable for efficient and accurate subject annotation of large-scale text data.