38 results about "Seed sample" patented technology

The invention discloses a seed vigor prediction model construction method based on spectral indexes and a detection system, and relates to a method and a system for rapidly and nondestructively detecting seed vigor by using a visible-near infrared spectrum technology. The problem that in the prior art, the requirement for accurate and quantitative evaluation of seed vigor in modern agriculture cannot be met is solved. The method comprises the following steps: firstly, acquiring spectral data and actual germination rate data of a seed sample, and carrying out abnormal value elimination and pretreatment on the spectral data; then extracting one-dimensional, two-dimensional and three-dimensional spectral indexes from the preprocessed spectral data, and screening out an optimal spectral index by using a correlation matrix method; and finally, by taking the optimal spectral index as input and the actual germination rate as output, establishing a seed activity prediction model based on the optimal spectral index. The prediction model is embedded in the system, and seed vigor prediction is achieved. The method is suitable for the fields of seed quality detection and grading, precise sowing, agricultural production and the like.

The utility model provides a layered sampling mechanism of a grain seed sampler, and belongs to the technical field of grain seed sampling. Comprising a sampling assembly, the sampling assembly is arranged in a supporting column, the sampling assembly comprises a baffle fixed in the supporting column, a sliding plate is arranged in the baffle, a supporting plate is arranged in the supporting column, a sliding groove is formed in the inner wall of the supporting plate, and the sliding groove is communicated with the sliding plate; a fixing rod is arranged at the top of the supporting plate, a guide groove is formed in the fixing rod, the supporting column accommodates the sampling assembly and is inserted into a grain pile and the sampling assembly through a bottom drill bit, the baffle is matched with the sliding plate, and under the action of the contraction assembly, the sampling assembly can stretch out of the sliding plate. Sliding is achieved through the guide groove of the fixing rod and the sliding groove of the supporting plate, resetting is assisted through the elastic piece, and therefore opening and closing of the material receiving open groove in the supporting column are controlled, and the effects of sampling in a layered mode and preventing grain leakage are achieved.

The invention relates to the technical field of artificial intelligence and software engineering, and discloses a small sample individual fairness testing method based on a domain adaptive flow model. Aiming at the problems of difficulty in obtaining target model training set data, low discrimination sample generation efficiency, insufficient sensitive attribute decoupling and the like in the existing black box fairness test, the invention provides the following technical scheme: firstly, constructing a gating residual domain adaptive flow model, and performing dynamic fusion on a frozen source domain projection matrix and a trainable low-rank adapter to obtain an adaptive flow model; domain adaptation of the flow model is realized under the condition of not depending on a target model training set; secondly, a two-stage anti-fact generation strategy is adopted, a seed sample set is constructed by utilizing a dynamic radius attenuation mechanism, and discriminated samples are efficiently expanded in combination with triple disturbance; finally, gradient orthogonal constraint is applied to the hidden space of the flow model, and decoupling control over the sensitive attributes and the semantic features is achieved. The black box fairness testing method can be used for black box fairness testing of classification models such as tables, texts and images, and has good practical significance and actual effect.

This invention discloses a molecular marker for rapid detection of highly homozygous fresh soybean germplasm. By carefully selecting specific editing sites, and then designing corresponding LAMP amplification primer sets and OSD probe sets based on these editing sites, the LAMP amplification primer sets and OSD probe sets designed in this invention can detect fresh soybean samples from different sources, verifying their specific recognition ability for autonomously edited improved germplasm. Wild-type OSD probes can also be used to detect seed samples from autonomous germplasm resources, verifying their accurate recognition ability for mixed wild-type seeds, ensuring that the seed purity detection error is ≤1%, meeting the requirements for production and planting.

According to the present disclosure, disclosed are a learning operating method based on a federated distillation, a learning operating server, and a learning operating terminal which calculate a local average logit by collecting data samples by the terminal, transmit the local average logit and seed samples to an uplink of a server, perform distillation of a global model based on the seed sample and the local average logit by the server to solve the problems of the privacy and communication overhead generated in the distributed network.

This invention discloses a deep neural network-based fuzzy testing method and system based on cross-modal generation and distribution awareness, belonging to the fields of software testing and artificial intelligence security. Addressing the problems of insufficient semantic diversity in generated test cases, uneven data distribution, and poor adaptability to high-resolution image datasets in existing testing methods, this invention uses clustering to select seed samples and extract abstract text embedding vectors. It then generates prompt words using a scene library, adjusts the generation quantity for each category according to a distribution awareness strategy, filters invalid data to obtain a valid test set, and inputs it into the model to statistically analyze error samples. This invention improves the diversity and effectiveness of test cases, increases test coverage and the ability to detect defects in few sample categories, and enhances model robustness.

The utility model provides a kind of agricultural product seed sample disinfection device.It relates to agricultural seed processing technical field, including box, jar, conveying belt, turnover assembly, feed hopper, cleaning assembly and support movement component;The jar is fixedly installed at the top of box, the bottom of jar is fixedly installed with discharge pipe, conveying belt is provided in the box, and the bottom end of discharge pipe is located directly above conveying belt, the rear side of box is equipped with the discharge port compatible with conveying belt, the turnover assembly is set on jar and extends to discharge cylinder, the feed hopper is fixedly installed at the top of jar, the cleaning assembly is set on box and jar, the support movement component is set on box.The utility model design is reasonable, easy and simple to operate, can realize efficient turnover, convenient discharge, comprehensive cleaning and flexible movement, to effectively improve the uniformity of seed disinfection, efficiency and equipment applicability.

A method and apparatus for generating an evaluation set for an intelligent marketing agent. The method includes: acquiring at least one seed sample; performing at least one sample generation based on the seed sample until a stopping condition is met; wherein, in one sample generation: generating candidate samples based on the seed sample and historical sample information; for each candidate sample, using multiple first-type agents to score the candidate sample, obtaining the scoring results and scoring reasons output by the first-type agents; using second-type agents to determine qualified samples that pass the screening and reference samples that fail the screening, and determining the reasons why the reference samples fail the screening, using the reasons as reference information for the reference samples; adding the reference samples and reference information to historical sample information, and adding qualified samples to a qualified sample set; in response to meeting the stopping condition, obtaining an evaluation set for the intelligent marketing agent based on the qualified sample set.

The invention discloses a rice seed germination detection method based on image recognition, and belongs to the field of image recognition, and the method comprises the steps: obtaining an original image sequence of a rice seed germination process based on a rice seed sample, and carrying out the preprocessing of the sequence, and obtaining an enhanced image sequence; processing the enhanced image sequence by adopting an image segmentation algorithm and an edge detection technology in sequence to obtain an independent feature map of a multi-layer structure; rice germination features are extracted from the independent feature map, the rice germination features comprise the coleoptile cracking degree, the endosperm density change and the total root length and the total root number, and the rice germination features are integrated into a comprehensive dynamic vector through a time sequence fusion method; calculating a consumption rate based on a coleoptile cracking degree and endosperm density change to obtain a correlation index of surface layer and middle layer change; and carrying out classification processing on the correlation indexes by adopting a convolutional neural network, judging the synchronism of multi-level change by integrating the total length and the total root number of the root system, and determining the overall health degree score of the germination state.

The invention provides a relation extraction method and device. The method comprises the following steps: determining a relation to be extracted and a seed sample; the seed sample comprises a labeled head entity, a labeled tail entity and a sentence containing the head entity and the tail entity; based on the to-be-extracted relation, at least one target character fragment is selected from the seed sample to serve as a template, and the target character fragment comprises a head entity or a tail entity; determining a target head entity and a target tail entity from the to-be-extracted text, and combining the template with the target head entity and the target tail entity to generate a cue word; performing similarity comparison on the cue word and the to-be-extracted text, and if the similarity is higher than a threshold value, adding a sentence containing a target head entity and a target tail entity into a seed sample to obtain an updated seed sample; and the relationship extraction step is repeated based on the updated seed sample and the to-be-extracted relationship until the to-be-extracted text does not contain the new entity pair, so that the manual annotation cost and errors are reduced, and the relationship extraction efficiency is improved.

This invention belongs to the technical field of salt-processed dodder seed quality evaluation, and discloses a method and system for salt-processed dodder seed quality evaluation based on multi-source information fusion. Traditional salt-processed dodder seed quality evaluation mainly relies on morphological identification and single-component content determination, which are highly subjective and cannot reflect the overall quality of the salt-processed dodder seed. This invention, based on salt-processed dodder seed samples with different processing degrees, obtains data on the appearance, color, taste profile, odor response, flavonoid content, and volatile component characteristics of the samples. Based on these five types of characteristic data, multi-source information fusion analysis is performed to construct a processing degree discrimination model. Finally, the processing degree discrimination model is used to determine the processing grade of the finished salt-processed dodder seed, thereby determining the quality grade of the medicinal material. This invention achieves comprehensive and multi-dimensional digital characterization of salt-processed dodder seed, deeply reflecting the impact of processing technology on the overall quality of the medicinal material, and overcoming the one-sidedness of single-component detection or subjective judgment.

The application provides a method for rapidly detecting seed germination rate, comprising: obtaining a plurality of seed samples with known germination rates, performing volatile organic compound detection on each sample to obtain volatile component data, and after data cleaning and arrangement, constructing a model of a machine learning algorithm of an integrated model, a linear model and a kernel method; using standardized feature data to establish an optimal prediction model of seed germination rate, which can be used for testing of unknown seed germination rate samples. Compared with a traditional manual statistical detection method, the detection period of the traditional method is compressed to within 2 hours, the detection efficiency is greatly improved, the manual germination and grain-by-grain counting operations are omitted, the labor input is significantly reduced, the human subjective error is reduced, the objectivity and repeatability of the detection result are greatly improved, and the error with the actual measurement value is within ± 5%. The method can be widely applied to seed quality detection scenes of various crops, and has good industry application and promotion value.

The invention discloses an annona squamosa seed vigor grading detection method, and relates to the technical field of annona squamosa seed vigor detection, and the method comprises the following steps: 1, collecting a seed sample; 2, seed pretreatment; 3, measuring the constant weight of the seeds; 4, measuring the water absorption rate of the seeds; 5, modeling after detection, and detecting and recording multiple data of the annona squamosa seeds through a TTC staining method, a conductivity determination test and a germination test; and step 6, vigor grading standard formulation. According to the method, multiple annona squamosa seed groups are adopted for detection, and multiple data of annona squamosa seeds detected through the water absorption rate, the TTC dyeing method, the conductivity determination test and the germination test are recorded, so that the method can be more convenient and more accurate when modeling comparison is performed on the data of the annona squamosa seed groups. Therefore, when the method is used for detecting and grading the activity of the annona squamosa seeds, the method can be more accurate, and the accuracy of the method during detection is improved.

The utility model discloses a seed detecting and sampling device which comprises a storage tank, a feed port and a discharge port are arranged at the upper end and the lower end of the storage tank, and a plurality of samplers are arranged on the side wall of the storage tank along the circumferential direction. According to the utility model, the samplers are arranged at all positions on the side wall of the storage tank, so that seeds at all area positions in the storage tank can be sampled randomly according to requirements, the sampled seeds can better reflect the quality of the seeds in the storage tank after being detected, the sampling is convenient and fast, and the sampling efficiency is improved. And the detected data after sampling is more accurate.

The invention discloses an oversampling and software defect prediction method based on space mapping and probability constraint. The method comprises the following steps: firstly, constructing a minority class probability score model by using dual-kernel density estimation, and realizing layered operation on a preprocessed defect sample; secondly, calculating the number of synthetic samples based on layered weight, performing three-dimensional space mapping and adaptive spherical domain nonlinear sampling on the seed samples, and reconstructing high-dimensional synthetic samples through inverse mapping; and finally, introducing manifold consistency and likelihood ratio dual probability constraints to perform quality verification on the samples, and triggering a radius feedback adjustment mechanism for unqualified samples until a balanced data set is constructed. And training the classifier by using the balanced data set to realize software defect prediction of an unknown sample. By selecting the high-quality seed samples and controlling the synthetic samples, the class imbalance problem is further processed from the data level, and the overall prediction performance of the model is remarkably improved.

The application discloses a zero sample generation method and device based on expert knowledge and Gaussian SMOTE, generates initial sample data by using a Gaussian noise addition method according to radar rules and expert knowledge accumulated through public data inquiry or other channels, further discriminates and screens the initial sample by using a K nearest neighbor algorithm to form seed samples, and finally expands the seed sample data by using an oversampling technology to form sample data suitable for intelligent algorithm training, so as to improve the generalization ability and applicability of the intelligent algorithm, improve the identification accuracy of the zero sample radar model, and solve the problem that the artificial intelligence identification method based on sample training cannot adapt to the situation that part of radar signals are less collected, and sample is missing in electronic reconnaissance and electromagnetic spectrum processing.

The invention discloses an incremental learning-oriented adversarial sample sustainability enhancement method and system, and mainly solves the problem that an existing adversarial sample loses efficacy due to domain drift caused by model updating in an incremental learning process. The scheme comprises: 1) training based on an incremental learning method to obtain an initial target model; 2) screening out a non-target sample without target category semantic tendency from the potential space as a seed sample, and performing feature enhancement; 3) extracting general text semantic features of a target category by using a pre-trained vision-language model, and constructing a semantic correction module in combination with gradient information of a target model; and 4) taking the seed sample and the enhanced sample as initial input, and generating an adversarial sample through iterative optimization under the guidance of a semantic correction module. According to the method, the attack sustainability of the adversarial sample after multiple rounds of updating of incremental learning can be remarkably improved, the generation efficiency and quality of the adversarial sample are improved, and the method can be used for security assessment and defense of a deep learning model.

The application provides a corn seed three-dimensional structure measurement method based on high-resolution tomography. A high-resolution tomography technology is used to scan a corn seed sample, and a tomography image of an internal structure of the seed is obtained; three-dimensional reconstruction is performed on the tomography image, and three-dimensional gray volume data of the internal structure of the seed is obtained; based on the three-dimensional gray volume data, an image segmentation technology is used to separate different structures in the seed, and a segmentation region of each structure of the seed is obtained; based on the segmentation region, three-dimensional characteristic parameters of each internal structure of the seed are calculated, and a seed vigor index is calculated according to the three-dimensional characteristic parameters. Lossless three-dimensional visualization and quantitative analysis of the internal structure of the corn seed can be realized, the volume, morphology and spatial distribution characteristics of each structure are accurately obtained, the biological activity of the seed is maintained, and technical support is provided for seed quality evaluation and breeding research.

The invention relates to the technical field of detection methods, in particular to a quality detection method for cynomorium songaricum seeds, which comprises the following steps: determining a seed monomer contour through image recognition and calculating a projected area mean value, obtaining a seed density mean value by combining multispectral analysis, and determining a winnowing wind speed based on the projected area mean value and the seed density mean value. According to the method, the clarity of the seed samples is quickly screened, and only the seed samples with the clarity reaching the standard are subjected to viability detection, so that the quality detection efficiency is improved, and the detection cost is reduced. The method specifically comprises the following steps: 1, determining image contours of a plurality of independent cynomorium songaricum seeds; 2, determining a plurality of uniformly distributed detection areas, and determining a projection area mean value of a target sample and a seed density mean value of the target sample; 3, determining a target wind speed value, obtaining two groups of samples, recording the samples as a seed sample and an impurity sample, and calculating a sample cleanliness value; and 4, carrying out viability detection on the qualified sample.

The invention provides a lotus seed producing area tracing method and system based on spectrum fusion. According to the method, the near infrared spectrum data and the hyperspectral image data of the lotus seed samples are processed through the combination of multiple preprocessing modes and multiple feature screening methods, classification results corresponding to different combinations are obtained after the data are input into the convolutional neural network, and the method combination corresponding to the classification result with the highest accuracy rate is selected. A lotus seed production place classification model is obtained after combination with a convolutional neural network (CNN), and in an actual test, the lotus seed production place classification model obtained through the method achieves 100% classification accuracy and is remarkably superior to a convolutional neural network model which only adopts a single spectral data source for classification prediction. According to the method, nondestructive and pollution-free testing can be achieved, the detection cost is low, the analysis speed is high, meanwhile, the near-infrared detection method has the advantages of being rapid, nondestructive, easy and convenient to operate and the like in practical application, and on-site rapid analysis can be achieved.

This invention relates to the field of motor fault detection technology, specifically a motor fault detection method and system based on vibration-current time-frequency co-characterization. The method synchronously acquires vibration signals, current signals, and operating condition reference information during motor operation to determine the fundamental current frequency and speed phase; identifies natural disturbance events and generates disturbance tags; uses these natural disturbance events as seed samples to extract current response and vibration response segments; performs disturbance synchronization alignment based on the disturbance tags, speed phase, and fundamental current frequency, and performs time-frequency folding and superposition to generate a natural disturbance folded development map; extracts fault dark pattern features from the map and outputs the motor fault detection results, applicable to online fault detection of motors under variable frequency speed control and load fluctuation conditions.

Described herein is a portable seed moisture content measuring system based on infrared reflection spectroscopy. The system comprises a portable SMC readout console and a portable seed threshing apparatus. The portable SMC readout console is a handheld device comprising a near-infrared sensor fixture. The sensor fixture comprises a base plate, one or more light emitting diodes (LEDs) on the base plate and oriented obliquely, The LEDs are configured to emit a beam of near infrared light to a seed sample and a photodetector disposed at a center of the base plate receives the reflected light. The portable seed threshing apparatus provides clean seeds for the SMC readout console.

The invention provides a technical blank space calibration and layout subject identification method based on suspected missed labeling patents, belongs to the technical field of patents, and solves the problems of technical blank misjudgment and the like caused by term diversity and indexing limitation in an existing patent analysis method. The method comprises the following steps: constructing a patent retrieval expression according to target technology description, performing retrieval, and constructing a large-field patent set; extracting high-purity seed samples from the large-field patent set, and constructing a high-purity seed set; modeling is carried out on the high-purity seed set from different technical perspectives, and a multi-dimensional behavior portrait is constructed; evaluating the to-be-evaluated patent based on the multi-dimensional behavior portrait, and generating a suspected label missing patent set; on the basis of the large-field patent set and the suspected missed labeling patent set, the technical blank area is calibrated, and a true blank and a false blank are distinguished; and identifying an underestimated layout main body and an emerging technical direction based on the large-field patent set and the suspected missed labeling patent set.

The present application relates to the field of traditional Chinese medicine and agricultural product quality analysis and intelligent detection technology, and discloses a method for intelligent identification of production area and prediction of active ingredients of snakegourd seed, which collects snakegourd seed samples from 11 main production areas in China, extracts morphological parameters, RGB, CIE XYZ, CIE Lab* color parameters and reconstructs visible spectral characteristics through machine vision, determines the contents of total triterpenoids, total polysaccharides, total proteins and 3,29-dibenzoyl triterpenoid, selects key features with VIP>1 by PLS-DA, constructs a firefly algorithm optimized RBF neural network model to realize identification of production area, and the accuracy of the test set reaches 87.39%, uses the maximum information coefficient to select core features, and combines BPNN to construct an active ingredient prediction model, and the R² of the test set of total triterpenoids, total polysaccharides and DBKT reaches 0.906, 0.936 and 0.935 respectively. The present application provides a method for traceability and nondestructive evaluation of the quality of snakegourd seed.

The application provides a rail transit X-ray security image contraband intelligent identification method and device, which can effectively expand high-quality training data, significantly improve small target and shielding contraband detection precision, and comprises the following steps: acquiring real X-ray contraband images in a rail transit security scene as seed samples to construct a text description library; generating contraband X-ray synthetic images through a generative model and performing domain adaptation adjustment through a conditional generative adversarial network; assigning sampling weights according to target size and shielding ratio, performing adaptive oversampling, and obtaining a balanced enhanced training data set; constructing a contraband detection network comprising a lightweight backbone network, a multi-scale feature pyramid, an attention-guided feature fusion module, a context enhancement module and an Anchor-Free detection head, and training to obtain a trained contraband detection model; and inputting a continuous X-ray image sequence to be detected into the trained contraband detection model to output a contraband detection result.

The invention discloses a method for detecting the content of isoquercitrin in a chingma abutilon seed sample and application of the method. A test solution and a reference solution are prepared respectively and injected into an ultra-high performance liquid chromatograph, and specific chromatograms of the reference solution and the test solution are compared; and calculating the content of the isoquercitrin in the chingma abutilon seed sample by utilizing a regression equation of the sample size and the peak area of the isoquercitrin. The method disclosed by the invention is good in specificity and durability, and a new analysis means can be provided for quality detection of a chingma abutilon seed sample; the specific component of isoquercitrin is taken as an index, so that the method is more reasonable and accurate; the method is applicable to detection of chingma abutilon seed medicinal materials, standard decoction, formula granules and other products containing chingma abutilon seeds. The method is simple to operate, good in reproducibility and stability, reliable in recovery rate, low in detection cost, high in detection efficiency and small in environmental pollution.

This invention discloses a multi-task classification method for rice seeds using a deep cross-disciplinary expert model. The method includes: preparing rice seed samples and acquiring their raw spectral image data; preprocessing the raw spectral image data of the rice seed samples to construct a dataset; constructing a deep cross-disciplinary expert model for classifying rice seed vigor and variety, and training the deep cross-disciplinary expert model using the dataset; acquiring the spectral image data of the rice seeds to be tested as input, and using the trained deep cross-disciplinary expert model to predict their seed vigor category and seed variety category. This invention emphasizes the integration of multi-task learning and near-infrared hyperspectral imaging to achieve efficient, accurate, and non-destructive simultaneous evaluation of seed vigor and variety, providing a practical solution for agricultural seed screening.

The invention discloses an oilseed rape variety nondestructive identification method, equipment, a medium and a product, and relates to the field of oilseed rape variety identification, the method comprises the following steps: obtaining oilseed rape seed samples of different varieties; acquiring a hyperspectral image of the rape seed sample, and correcting the hyperspectral image; performing spectrum extraction on the corrected hyperspectral image; according to the extracted spectrum, spectral feature extraction is carried out by adopting a feature selection competitive self-adaptive reweighted sampling method; determining a classification model by adopting a support vector machine according to the extracted spectral features; and carrying out rape variety identification by using the classification model. The accuracy and speed of rape seed identification can be improved.

The invention relates to the technical field of image recognition, and discloses a seed lesion grain recognition method and a seed screening system.The method comprises the steps that foreground perception fusion is carried out on seed sample images to obtain standardized images, and channel differentiation enhancement is carried out to obtain enhanced images; analyzing microcosmic and macroscopic visual features and carrying out double-flow coupling to obtain comprehensive feature expression; positioning a lesion area of the seed according to collaborative abnormal performance of texture features and spectral features in the comprehensive feature expression, and performing morphological closing on the lesion area to obtain a lesion area boundary; according to the morphological characteristics and the coverage range of the lesion area boundary, visual mode discrimination is carried out on the health state of the seeds to obtain a recognition conclusion; sending a state identifier in the identification conclusion to a sorting execution terminal to complete seed screening; according to the invention, the seed lesion particle identification efficiency can be improved.

The utility model provides a crop seed sample divider, which relates to the technical field of seed sample division and comprises a base, the base comprises a bottom plate, and a plurality of mounting holes are formed in the bottom plate. According to the crop seed sample divider, the motor drives the material dividing barrel to rotate so that seeds can be thrown into the multiple material dividing pipes, meanwhile, the seeds are distributed to the multiple groups of material dividing pipes, when the number of the distributed seeds in each group needs to be adjusted, a limiting rod is pulled through a pull block, a tension spring can deform, and after the limiting rod is separated from limiting of a limiting hole, a movable ring can slide at the moment; a movable ring moves on the material distributing pipe, then an L-shaped rod and a baffle are driven by the movable ring to move, after the baffle is moved to a proper position, a pull block is loosened, a limiting rod is clamped into another limiting hole again under the elastic force action of a tension spring, and at the moment, a seed storage space formed by the movable ring and the material distributing pipe changes; and the quantity of the distributed seeds is adjusted.