39results about How to "Have the ability to identify" patented technology

The invention discloses a human face verification method based on bilinear united convolutional nerve network. The human face verification method comprises steps of 1) using a human face image which is prepared in advance to perform convolutional nerve network (CNN) training, 2) using the human face image which is in a training set to perform bilinear CNN fine tuning, 3) inputting a human face image to be verified, segmenting the two images, extracting united characteristics outputted by the bilinear CNN, and 4) making an obtained vector go through self-encoding network training to obtain a final verification result. The human face verification method is based on the bilinear CNN, replaces two repeated inputs of an original bilinear nerve network with different human face verification input images, and brings forward a human face verification description factor. The human face description factor has robustness to illumination, shielding and posture change. Furthermore, the characteristic extracted by the bilinear CNN has a smaller dimensionality than the characteristic dimensionality of a common CNN fully connected layer, which reduces number of parameters, makes follow-up deep belief network training simple and improves accuracy of human face verification.

The correcting method for dynamic measured position of photoelectronic width measurer is suitable for real-time correction of measurement error caused by the axial runout and inclination of the measured object during linear measurement of photoelectronic width measurer. The method includes extracting axial position information from the width information obtained with the detector so as to make the width measurer possess capacity of recognizing the object position in the optical axis direction and to constitute one integral width measuring and ranging measurement system; calibrating the video camera through experiment; and establishing the target axial runout and inclination correcting relationship with the correcting model, so that the photoelectronic width measurer can correct measurement error timely based on the position information during dynamic measurement.

The invention discloses a satellite image ship model recognition method under a small sample condition, and the method comprises the steps of obtaining a to-be-recognized ship remote sensing image of a target region, and carrying out the preprocessing of the to-be-recognized ship remote sensing image; inputting the preprocessed ship remote sensing image into a trained few-sample learning classifier and C SVM classifiers; and outputting a ship model corresponding to the ship remote sensing image subjected to multi-model fusion through an integration strategy, wherein the few-sample learning classifier and the C SVM classifiers measure distance information between the to-be-identified image and different types of targets in the small sample support set by using metric learning on the basis of a metric learning technology, so that the data feature extraction and expression capability under a small sample condition is improved, and the target identification precision under the small sample condition is improved. And on the other hand, an ensemble learning method is adopted, a classic machine learning technology is combined, an ensemble strategy is designed, ship target model stable recognition under the small sample condition is achieved, and the ship target model recognition capability is achieved.

The invention discloses a gesture data set acquisition method for a YOLO network, and a gesture recognition method and device. The acquisition method comprises the steps that the frame including the training gesture is acquired, the image acquisition window frame is saved as the training gesture image, the training gesture image background is filtered by using a simple threshold skin color segmentation algorithm based on the YCbCr color space so as to obtain the training gesture image of diverse backgrounds, and the training gesture image of diverse backgrounds is applied to form the gesture data set. The gesture recognition method comprises the steps that YOLO network training and gesture recognition are performed by using the obtained gesture data set. The device comprises a memory and aprocessor. The YOLO network is enabled to have the recognition model for the gesture in the image or the frame, and the recognition process has high noise suppression capacity for the face, the skinlike wall and other complex backgrounds existing in the image. The gesture data set acquisition method for the YOLO network, and the gesture recognition method and device are applied to the technicalfield of image recognition processing.

The present invention discloses a radar target recognition method based on tag consistent dictionary learning, and belongs to the field of radar. Sparse representation of signals is realized by constructing a complete dictionary with a discriminating ability and sparse coding for sparse signals. The method comprises the main flows of: constructing a LC-KSVD dictionary learning model of a high-resolution range image; initializing relevant parameters in the model; using a K-SVD algorithm to solve the optimal solution of the LC-KSVD dictionary learning model; normalizing the dictionary D and thelinear classifier matrix W again; and finally, according to the D and the W, determining the category of the test sample. According to the method disclosed by the present invention, the learned complete dictionary has the discriminating ability, so that not only the sparse representation of the signals can be better realized, but also the method can be used for classification.

The invention discloses a remote sensing image target extraction system and method based on deep learning, and the system comprises: a backbone network module which is used for carrying out the downsampling of an original image for many times, and obtaining a first low-level feature, a second low-level feature, a third low-level feature and a fourth low-level feature; the discrimination context perception feature extraction module that is used for obtaining a multi-scale context feature difference fusion result according to the fourth low-level feature; the first up-sampling module that is used for obtaining a first advanced feature according to a multi-scale context feature difference fusion result; the first refining decoder module that is used for fusing and up-sampling the third low-level feature and the first high-level feature to obtain a second high-level feature; and the second refining decoder module that is used for fusing and up-sampling the second low-level feature and thesecond high-level feature result to obtain a third high-level feature. The remote sensing image target extraction system and the remote sensing image target extraction method can enhance the identification capability of background and target features, and have good target extraction capability.

The invention discloses a preparation method for a solid phase microextraction recognizable adsorption coating, comprising the steps that a support stainless steel wire is pretreated, carbon nanotubes (CNTs) are activated, a SiO2 sol solution is prepared, CNTs-MIP (molecularly imprinted polymer) polymer dope is prepared, a fiber tip of the treated stainless steel wire is vertically soaked into the SiO2 sol solution and is irradiated under an ultraviolet lamp for 30 min after being taken out, and a SiO2 coating is prepared; the fiber tip is then soaked into the CNTs-MIP polymer dope and is irradiated under the ultraviolet lamp for 30 min after being taken out, and a CNTs-MIP coating is prepared; the operations are carried repeatedly and alternately until the thickness of the coatings reaches 20-50 mu.m; and the prepared coatings are placed at a gas chromatography injection port and aged in the nitrogen condition at the temperature of 300 DEG C until no template molecules are detected in a chromatogram, namely an SPME recognizable adsorption coating of SiO2 / CNTs-MIP, having good recognition and extracting power towards a target substance, is obtained.

The invention provides an eddy current recognition composite material and equipment on the interface thereof, belonging to the technical field of metal material recognition. An adjustable signal generation module generates electric signals with frequency being adjustable continuously; furthermore, the signals are respectively transmitted to the eddy current probe of a standard part and the eddy current probe of a part to be measured, and the eddy current probe of the standard part and the eddy current probe of the part to be measured are connected with a probe lifting mechanism; two probes convert the input electric signals into magnetic signals, emits the magnetic signals to the surface of the composite lamellar material and receives the returned signals; a signal processing module respectively receives the signals output by the two probes; a subsequent execution mechanism is controlled by two signals; an optical-coupling switch detects the position of the workpiece; a micro-processing module controls the adjustable signal generation module to send signals to the eddy current probe of the standard part and the eddy current probe of the part to be measured and controls the execution mechanism to start operations. The composite material and the equipment have recognition function for the composite lamellar material and can position for the recognition of the different material interfaces of the composite material; the operation is simple and the application is wide.

The invention relates to an unmanned aerial vehicle management method, a terminal device and a memory medium. The method comprises the following steps of S100, inputting frequency hopping sequences after frequency pairing is carried out on licensed remote controllers and unmanned aerial vehicles; S200, recording radio frequency signals within a frequency hopping sequence period in a fixed radio frequency band through utilization of a spectrum monitoring device, determining a radio frequency signal group of the remote controllers by removing interference signals, and calculating the frequency hopping sequence of each remote controller in the radio frequency signal group; and S300, comparing the calculated frequency hopping sequence of each remote controller with the input frequency hoppingsequences of the licensed remote controllers in sequence, and finding out the frequency hopping sequences of unlicensed remote controllers. According to the method, the terminal device and the memorymedium, through analysis, extraction, judgment and interference of the radio frequency signals of the remote controllers of the unmanned aerial vehicles, the unmanned aerial vehicles are precisely managed.

The invention discloses automatic pasting and scanning equipment of embryos bar codes. The equipment comprises a PLC, a frame, and a bar code conveying and scanning unit and a bar code pasting unit arranged on the frame; the bar code conveying and scanning unit includes a feeding mechanism, a label stripping mechanism, a bar code storage table, a scanning module, a label detecting mechanism, a collecting mechanism and a guide mechanism; the bar code pasting unit includes a rotating cylinder, an extension cylinder fixedly connected with a cylinder rod of the rotating cylinder, and a bar code picking / releasing part fixedly connected with a cylinder rod of the extension cylinder; the collecting mechanism is connected with a driving motor for driving the collecting mechanism to collect materials; and the scanning module, the label detecting mechanism, the rotating cylinder, the extension cylinder, the bar code picking / releasing part and the driving motor are electrically connected with the PLC. The equipment has the beneficial effects of possession of automatic identifying and control capacity, guarantee accurate embryos production, possession of material identifying capacity, verification of material information and possession of a control operation step.

The invention discloses base-Schiff base derivatives. A structural general formula of the base-Schiff base derivatives is as shown in a formula [I] as shown in the specification, and the base-Schiff base derivatives are produced by subjecting raw materials of p-bromoaniline, paraformaldehyde, n-butyllithium, p-aminobenzoic acid, substituted o-phenylenediamine and the like to a condensation reaction. The base-Schiff base derivatives fluoresce, emit blue light, and have obvious aggregation-induced emission and solid-state emission, part of compounds have capability of recognizing Al3+, and two compounds have good effects of inhibiting one or two kinds of triple-negative breast cancer cells and liver caner cells, and have small toxicity to normal cells; and three compounds have certain degrees of effects of inhibiting one or more kinds of non-small cell lung cancer cells, triple-negative breast cancer cells and liver cancer cells.

The invention provides a benzothiadiazole-TB compound as well as a synthesis method and application thereof; the structural general formula of the benzothiadiazole-TB compound is represented by a formula A, wherein R is one of hydrogen, alkoxy, hydroxyl and alkyl. The benzothiadiazole-TB compound has characteristics of large Stokes displacement, excellent luminescence property, excellent solid state luminescence, and potential of becoming an excellent OLED material; the benzothiadiazole-TB compound has a wide pH application range and can be applied to a human physiological environment; the benzothiadiazole-TB compound has effective and good response to viscosity, and has the possibility of becoming a viscosity response fluorescent probe; the benzothiadiazole-TB compound has recognition capability on Fe<3+> and is expected to become an excellent Fe<3+> fluorescent probe; and the benzothiadiazole-TB compound has a good PDT effect, can perform dyeing imaging on A549 cells, expands the variety of photosensitizers, and provides a new thought for synthesizing imaging-guided PDT photosensitizers.

The invention discloses a ship classification and identification method and system based on track features and deep neural network MobileNet migration training. The method comprises the following steps of: firstly, mapping characteristics such as speed, course and acceleration obtained by mining track data to an RGB (Red, Green, Blue) color space, and converting the characteristics into track characteristic image data; then, migrating pre-training weights of a depth neural network MobileNet and an Image Net of the depth neural network MobileNet, and training the network by using track feature image data generated before; and finally obtaining a ship type identification model to realize ship classification and identification. The method has obvious advantages in recognition accuracy and speed, and can be effectively applied to classification and recognition of ship targets.

The invention discloses a D-benzothiadiazole-TB(-D) derivative as well as a synthesis method and application thereof. The structural formula of the D-benzothiadiazole-TB(-D) derivative is shown as a formula I or II. The D-benzothiadiazole-TB(-D) derivative has large Stokes displacement, shows excellent luminescence property, has excellent solid luminescence and has the potential of becoming an excellent OLED (Organic Light Emitting Diode) material. The D-benzothiadiazole-TB(-D) derivative has a wide pH application range and can be applied to a human physiological environment; the D-benzothiadiazole-TB(-D) derivative has effective and good response to viscosity, and has the possibility of becoming a viscosity response fluorescent probe; the D-benzothiadiazole-TB(-D) derivative has recognition capability on Fe<3+> and is expected to become an excellent Fe<3+> fluorescent probe; and the D-benzothiadiazole-TB(-D) derivative has a good PDT effect, can perform dyeing imaging on A549 cells, and expands the variety of photosensitizers.

The invention provides an eddy current recognition composite material and equipment on the interface thereof, belonging to the technical field of metal material recognition. An adjustable signal generation module generates electric signals with frequency being adjustable continuously; furthermore, the signals are respectively transmitted to the eddy current probe of a standard part and the eddy current probe of a part to be measured, and the eddy current probe of the standard part and the eddy current probe of the part to be measured are connected with a probe lifting mechanism; two probes convert the input electric signals into magnetic signals, emits the magnetic signals to the surface of the composite lamellar material and receives the returned signals; a signal processing module respectively receives the signals output by the two probes; a subsequent execution mechanism is controlled by two signals; an optical-coupling switch detects the position of the workpiece; a micro-processing module controls the adjustable signal generation module to send signals to the eddy current probe of the standard part and the eddy current probe of the part to be measured and controls the execution mechanism to start operations. The composite material and the equipment have recognition function for the composite lamellar material and can position for the recognition of the different material interfaces of the composite material; the operation is simple and the application is wide.

The invention discloses a complex amplification kit for supplementing STR site for difficult detection materials and application thereof. The kit comprises 17 pairs of specific amplification primers of STR loci, wherein the STR loci are Amelogenin, D3S1358, D6S477, D1S1656, D19S253, D2S441, D6S1043, D15S659, D3S3045, Penta E, D8S1132, Penta D, D10S1435, D22S1045, D10S1248, D12S391 and DYS391. Thekit has the advantages of rapid and efficient detection of the difficult detection materials such as trace, degradation and inhibitor-containing materials in a field case; and meanwhile, the kit has high efficiency for amplification of the difficult case detection materials, and the optimized locus information can supplement the existing case database, shorten the case investigation time and lockthe suspect.

The invention relates to a video monitoring equipment information acquisition control implementation method which comprises the following steps: detecting the survivability of an IP address, and judging whether the IP address has survivability or not; otherwise, exiting the step; analyzing protocol flow through a PCAP technology and a deep packet inspection technology, and identifying asset information to obtain dpi_data; sending a simulated SSDP request to the IP address, and extracting a specified key character string for describing equipment parameters by analyzing the xml file to obtain upnp_data; carrying out model data matching through the cgi_info script and the hiki_info script; and obtaining the type of the video monitoring equipment through the obtained information so as to obtain technical parameter information. By adopting the video monitoring equipment information acquisition control implementation method, through the technical scheme of data communication protocol analysis and an active and passive mode, the identification accuracy and the identification comprehensiveness of the technical parameter information are improved. Through a collision algorithm and model fitting, a digital model is established for unknown equipment, and complete dependence of a traditional method on a fingerprint database is eliminated.

The invention provides an implementation method and system for user dressing detection, a medium and a terminal. The method comprises the following steps: acquiring a work clothes image set and a to-be-detected user image; performing feature extraction on each work clothes image in the work clothes image set to obtain a first feature vector corresponding to each work clothes image; based on the to-be-detected user image, obtaining a second feature vector corresponding to the to-be-detected user dressing on the to-be-detected user image; calculating the similarity between the second feature vector and each first feature vector so as to detect the dressing of the to-be-detected user according to the similarity; whether the user wears the work clothes or not is judged by calculating the similarity between the dressing of the to-be-detected user and the work clothes image, so that the situation that an algorithm network needs to be retrained along with updating or expansion of the work clothes can be effectively avoided, and the method is suitable for various work clothes styles and has industry application adaptability and batch popularization capacity; and a unified algorithm networkcan be utilized to adapt to various application scenes.

The invention discloses automatic pasting and scanning equipment of embryos bar codes. The equipment comprises a PLC, a frame, and a bar code conveying and scanning unit and a bar code pasting unit arranged on the frame; the bar code conveying and scanning unit includes a feeding mechanism, a label stripping mechanism, a bar code storage table, a scanning module, a label detecting mechanism, a collecting mechanism and a guide mechanism; the bar code pasting unit includes a rotating cylinder, an extension cylinder fixedly connected with a cylinder rod of the rotating cylinder, and a bar code picking / releasing part fixedly connected with a cylinder rod of the extension cylinder; the collecting mechanism is connected with a driving motor for driving the collecting mechanism to collect materials; and the scanning module, the label detecting mechanism, the rotating cylinder, the extension cylinder, the bar code picking / releasing part and the driving motor are electrically connected with the PLC. The equipment has the beneficial effects of possession of automatic identifying and control capacity, guarantee accurate embryos production, possession of material identifying capacity, verification of material information and possession of a control operation step.

The invention involves a video monitoring equipment information collection control implementation method, including the survival of the IP address, and determining whether the IP address is surviving; otherwise, the withdrawal steps; analyze the traffic of the protocol through PCAP technology and deep package detection technology, identify assets of assets, identify assetsInformation, get DPI_DATA data; send an analog SDP request to the IP address. By parsing the XML file, the specified key string is extracted to describe the device parameters to obtain the UPNP_DATA data;The type of information obtaining video monitoring equipment is based on technical parameter information.The implementation of a video surveillance equipment information collection control implementation method of the present invention is adopted. Through the technical solution of data communication protocol analysis and active passive mode, the accuracy of technical parameter information is improved and comprehensive.Through the collision algorithm and model fit, the digital model is established on the unknown device, and the traditional method is relied on the traditional method to rely on the fingerprint library.

The invention relates to a tripodal hexaurea neutral ligand as well as a preparation method and application thereof, and belongs to the technical field of high polymer materials. A triamine bridging group is used as a framework, a repeatable o-phenylenediurea structure is used as a basic unit, and a tripodal hexaurea neutral ligand is formed through tail end substituents of-CnH2n + 1,-PhCnH2n + 1 or-PhOCnH2n + 1. The preparation method comprises the following steps: directly reacting isocyanate with a compound I to obtain a tripodal hexaurea neutral ligand; or the amine substance is converted into carbamate, and then the carbamate reacts with the compound I to obtain the tripodal hexaurea neutral ligand. The ligand can be used for identifying anions, and especially has high extraction efficiency and high selectivity on sulfate radicals.

The invention discloses a radar target recognition method based on label-consistent dictionary learning, which belongs to the field of radar. In particular, the sparse representation of the signal is realized by constructing a complete dictionary with discriminative ability and sparse coding of the sparse signal. The main process of the method: first, construct the LC-KSVD dictionary learning model of high-resolution range images; then initialize the relevant parameters in the model; then use the K-SVD algorithm to solve the optimal solution of the LC-KSVD dictionary learning model; next, Then normalize the dictionary D and the linear classifier matrix W; finally, use D and W to determine the category of the test sample. The method of the invention can make the learned complete dictionary have discrimination ability, not only can better realize the sparse representation of signals, but also can be used for classification.

A class of base-Schiff base derivatives, whose general structural formula is shown in the following formula [I], which is formed by condensation of raw materials such as p-bromoaniline, paraformaldehyde, n-butyllithium, p-aminobenzoic acid, substituted o-phenylenediamine, etc. The reaction is made. The fluorescence of base-Schiff base derivatives of the present invention is blue luminescence, has obvious aggregation-induced luminescence and solid-state luminescence, and some compounds are sensitive to Al 3+ With recognition ability, the two compounds have a good inhibitory effect on one or both of triple-negative breast cancer cells and liver cancer cells, and are less toxic to normal cells; the three compounds are effective against non-small cell lung cancer cells, triple-negative breast cancer cells One or more of the three types of cells and lung cancer cells have a certain degree of inhibitory effect.

The invention discloses a nucleic acid aptamer with high affinity to okadaic acid and dinophysistoxin, which comprises two sections of split type aptamers DSP-1 and DSP-34, and the nucleotide sequences of the two sections of split type aptamers DSP-1 and DSP-34 are as shown in SEQ ID NO.1 and SEQ ID NO.2. The invention further discloses another nucleic acid aptamer, namely, DSP-134, which has high affinity to okadaic acid and dinophysistoxin, and the nucleotide sequence of the nucleic acid aptamer is shown as SEQ ID NO.5. The Kd value of the aptamer DSP-1 and the aptamer DSP-34 combined with OA is 11.6 nM, and the Kd value of the aptamer DSP-1 and the aptamer DSP-34 combined with DTX is 43.7 nM; the Kd value of the aptamer DSP-134 combined with OA is 133nM, and the Kd value of the aptamer DSP-134 combined with DTX is 228nM; and the nucleic acid aptamer has no recognition capability on marine toxins such as saxitoxin, neosaxitoxin, gonyautoxin and the like. According to prediction of a secondary structure of an original aptamer, a transformation mode of firstly truncating and then cracking is adopted, so that the transformed aptamer has better affinity to a target and has recognition capability to both OA and DTX.

The invention discloses a face recognition reminding system based on a convolutional neural network. The system comprises a raspberry pi control module, a voice output module and a video acquisition module. The raspberry Pi control module is connected with a camera of the video acquisition module through a CSI port. Starting and working of the camera are controlled by the CSI port. The raspberry Pi control module is connected with a control pin and a power supply pin of the voice output module through a voice output control signal port and a power supply port of GPIO. According to the invention, the convolutional neural network is adopted to carry out face recognition and prompt, the face detection is not influenced by the position of the face, and the recognition accuracy is good.

The invention belongs to the technical field of dynamic lighting, in particular to a vision capture illuminating system based on the internet of things and a control method of the illuminating system. The vision capture illuminating system is formed by connecting multiple illuminating devices. Each illuminating device comprises a frog eye bionic vision capturing unit, an internet of things coordinating unit and an illuminating device control unit, wherein the frog eye bionic vision capturing unit is connected with the internet of things coordinating unit, and the internet of things coordinating unit is connected with the illuminating device control unit. The multiple internet of things coordinating units are connected through a wireless multi-hop networking protocol, so that an intelligent coordination center is formed. The control method comprises the steps that the size, the movement direction, the movement speed and the acceleration of a moving object in a view field of the system are measured by the capturing units and sent to the internet of things coordinating units, the internet of things coordinating units send corresponding instructions to the control units of all the illuminating devices, in this way, it is guaranteed that dynamic illuminating support effect is achieved in the advancing direction of the moving object, and a sufficient view field range and sufficient light brightness of the illuminating system and the response time in emergency are guaranteed.

The invention relates to an unmanned aerial vehicle management method, a terminal device and a memory medium. The method comprises the following steps of S100, inputting frequency hopping sequences after frequency pairing is carried out on licensed remote controllers and unmanned aerial vehicles; S200, recording radio frequency signals within a frequency hopping sequence period in a fixed radio frequency band through utilization of a spectrum monitoring device, determining a radio frequency signal group of the remote controllers by removing interference signals, and calculating the frequency hopping sequence of each remote controller in the radio frequency signal group; and S300, comparing the calculated frequency hopping sequence of each remote controller with the input frequency hoppingsequences of the licensed remote controllers in sequence, and finding out the frequency hopping sequences of unlicensed remote controllers. According to the method, the terminal device and the memorymedium, through analysis, extraction, judgment and interference of the radio frequency signals of the remote controllers of the unmanned aerial vehicles, the unmanned aerial vehicles are precisely managed.

The invention discloses a method for locking and extracting a target sound source, which comprises the following steps of: extracting a plurality of sound sources through a multichannel BSS algorithmAUX-IVA (independent vector analysis based on an auxiliary function), and performing sound source selection through self-adaptive target sound source locking based on SNR (Signal to Noise Ratio) and voice recognition system feedback. On the basis, a pure interference sound source section is judged through spatial information opposite to different sound sources in sound source separation, and elimination is carried out. Finally, VAD endpoint detection based on DNN is used. A final target sound source effective voice segment is extracted through the algorithm, and feedback of the voice recognition system also acts on the VAD algorithm. According to the AUX-IVA algorithm based on the time window, the number of convergence iterations of the overall algorithm is small, the algorithm is lighter,and space information is further used for separating and extracting an interference source and a target source.