34results about How to "Help capture" patented technology

The invention discloses a multi-scale CNN-BiLSTM non-coding RNA interaction relationship prediction method with introducing attention, and belongs to the field of bioinformatics and deep learning. Themethod comprises the following steps: (1) proposing a coding mode k-mers suitable for a gene sequence; (2) using a multi-scale convolution kernel for replacing a single-scale convolution kernel, so that topic features with different lengths between sequences are captured, the feature diversity is enriched, and the model prediction performance is improved; carrying out down-sampling on each convolved feature map by using a plurality of pooling windows with different scales, so as to avoid ignoring potential effective information; (3) fusing a BiLSTM model on the basis of the CNN, so that long-distance information dependence between sequences can be better processed, and feature information is fully learned; and (4) introducing an attention mechanism, and distributing different weights to different words in the text vector by using the attention mechanism to distinguish the importance of the information, so that the attention mechanism pays more attention to the key information, and thepurpose of enhancing learning is achieved.

The invention discloses a vehicle-mounted traveling record management system used for a large van. The vehicle-mounted traveling record management system comprises an image monitoring module, a vehicle condition judgment module, a GPS module, an outer receiving module, a wireless communication module and a driving prompting module. By the adoption of the vehicle-mounted traveling record management system used for the large van, multiple cameras are arranged for panoramically monitoring the front side, the rear side, the left side and the right side of the van, the problem that a vision dead zone is generated due to the high driving position of the van is solved, the traffic safety coefficient can be better increased, and parking is convenient. By the adoption of the wireless communication module of the system, the van can be located and checked in real time by a driver and a van management center conveniently.

The invention discloses a traffic speed prediction method. The method comprises the following steps of acquiring a data set including road network information and vehicle speed information; extracting spatial correlation of K-order adjacent road sections in the road network by using a graph attention mechanism, and outputting a vehicle speed feature matrix V (K) of each road section; and using a time convolution network to learn a time rule of V(K), outputting a time sequence feature matrix, using a multi-head self-attention mechanism to learn a coupling relation between the feature matrixes, and outputting the coupling relation. According to the invention, the graph attention mechanism and the time convolution network are used to extract space dependence and time dependence; and the multi-head self-attention mechanism is introduced to extract a space-time coupling effect, and the multi-head self-attention mechanism can learn information from different representation subspaces at different positions so that the time information of a traffic flow can be captured, and higher prediction precision can be realized.

The invention discloses an unmanned vehicle camera system. The system comprises monocular cameras, a camera array, pan heads, a data processing system and an unmanned vehicle. The pan heads are installed on the two sides and the middle of the unmanned vehicle and used for adjusting the shooting direction of the corresponding camera and improving the stability of the camera. The monocular cameras are installed on the pan heads located on the two sides to form the wide-baseline camera array. The camera array is installed on the middle pan head and used for obtaining the image information of thesurrounding environment of the unmanned vehicle and the tracking target. The data processing system is installed in the unmanned vehicle and connected with the monocular cameras, the camera array andthe pan heads and used for controlling the pan heads and obtaining image information captured by the cameras and then processing the image information. The system has the combination of the short-baseline camera array and the wide-baseline camera array to generate multiple parallax and depth maps so as to obtain the accurate depth information of the surrounding environment of the unmanned vehicle.The short-baseline camera array is additionally arranged for realizing synthetic aperture imaging for tracking the occluded object, i.e. de-occlusion is realized, and the high-resolution and high-dynamic fusion image can be obtained.

The invention discloses a remote sensing image semantic segmentation method based on parallel cavity convolution, and relates to the technical field of remote sensing images, and the method comprises the following steps: obtaining a high-resolution remote sensing image in advance, and carrying out the slicing, normalization and standardization of the high-resolution remote sensing image, and obtaining a source high-resolution remote sensing image; initializing a low-layer network of a feature extraction network resnet101 based on a resnet101 parameter pre-trained on an ImageNet, constructing a parallel cavity convolutional network, and extracting a shallow-layer feature of a source high-resolution remote sensing image; inputting the shallow layer features into a parallel dilated convolutional network to obtain multi-scale information, and fusing the multi-scale information; and fusing the fused features with the shallow features again, and repairing image-level information by using a full-connection conditional random field to obtain a semantic segmentation result. Under the condition that extra parameters are not increased, the receptive field of convolution is expanded, and compared with standard convolution achieving the same receptive field, the parallel cavity convolution method can save more video memories.

According to the centrifugal air purifier, a cylindrical inner cavity is formed in a shell, an air inlet is formed in one end of the cylindrical inner cavity, an air outlet is formed in the other end of the cylindrical inner cavity, a centrifugal fan is arranged in the cylindrical inner cavity, the air inlet side of the centrifugal fan is right opposite to the air inlet, and the air outlet side of the centrifugal fan is distributed on the peripheral wall of the centrifugal fan; the air outlet side corresponds to the inner peripheral wall of the cylindrical inner cavity; the photocatalyst carrier is arranged on the surface of the inner wall of the barrel-shaped inner cavity, and the ultraviolet light source is arranged in the barrel-shaped inner cavity and used for irradiating the photocatalyst carrier. By means of the centrifugal fan and the cylindrical inner cavity, air rotates at a high speed to generate centrifugal force, pollutants carried in the air are compressed on the surface of the photocatalyst carrier on the inner wall of the device, and the decomposition and disinfection efficiency can be improved; under the action of frictional resistance of the surface of the photocatalyst carrier, a continuous rolling forward vortex is generated, so that more gaseous pollutants can be captured by the photocatalyst, and the purification efficiency is improved.

The invention relates to a remote sensing image conversion map migration method based on a semi-supervised generative adversarial network. The method comprises the following steps: constructing a semi-supervised map generation model; performing supervised training on a generator and a discriminator according to the paired training data set and a preset loss function, adding transformation consistency regularization loss in a pre-training generator, performing channel width expansion on the pre-training generator, and adding expansion loss and channel loss for training to obtain an expansion generator; and initializing an unsupervised discriminator by using the weight of the pre-training discriminator, performing adversarial training on the trained extension generator, the unsupervised discriminator, the pre-training generator and the pre-training discriminator according to preset unsupervised adversarial loss, and performing map conversion on an input remote sensing image according to a trained semi-supervised map generation model. By adopting the method, the geometric accuracy and visual quality of the map generated by the remote sensing image can be improved.

The invention provides a method for preparing isosteviol through catalytic hydrolysis of stevioside by chlorinated aromatic sulfonic resin, which belongs to the technical field of organic synthesis of organic compounds. According to the invention, a chlorinated aromatic sulfonic ion exchange macroporous resin packed bed is used for continuous catalytic hydrolysis of stevioside so as to prepare isosteviol, and a step-by-step cooling and circulation manner is adopted to remove products in time and to accelerate reaction. The method provided by the invention has the advantages of high yield, simple operation, environment friendliness, good stability performance of the catalyst, etc. The strong chlorinated aromatic sulfonic ion exchange macroporous resin can maintain 97% of catalytic activity after repeatedly used 10 times. Yield of isosteviol can reach more than 95%, and purity of isosteviol is more than 98% after simple recrystallization.

The invention discloses an oil absorption and impurity removal type resin adsorption hemisphere, and belongs to the technical field of oil absorption resin. A plurality of oil stain adsorbers positioned below resin oil absorption cotton are used for upwards adsorbing oil-containing suspended matters; a resin adsorption rod positioned below a water surface is used for upwards adsorbing the oil-containing suspended matters; non-oil particles in the oil-containing suspended matters are separated from oil dirt by utilizing the matching of an outer oleophobic adsorption layer and a magnetic oleophobic adsorption layer arranged at the outer side part of the top end of the resin adsorption rod; the oil dirt is guided into the resin oil absorption cotton along with a high oil absorption cotton core; impurities such as the non-oil particles are guided to a scum adsorption body through the magnetic oleophobic adsorption layer and the outer oleophobic adsorption layer; and when scum collection balls on the scum adsorption body are adsorbed to a certain weight, the scum collection balls are separated from a guide adsorption rod and fall onto a adsorption fiber layer, so that the follow-up collection of the non-oil particles by technicians is facilitated, and after the impurities in the oil dirt adsorbed and collected in the resin oil absorption cotton are removed, subsequent recovery is more convenient.

The invention relates to a preparation method and application of a bimetallic nano enzyme, and the preparation method comprises the following steps: (a) preparing a hexagonal metal organic framework material ZIF-67 by adopting a solvent method; (b) synthesizing nickel cobaltate (NiCo2O4) by adopting an ion replacement method; and (c) preparing the NiCo2O4-coated PVP bimetallic nano-enzyme through a one-step template self-assembly method, so that the activity of superoxide dismutase (SOD enzyme) and the activity of catalase (CAT enzyme) can be well simulated, and. OH can be effectively removed. The preparation method disclosed by the invention has the beneficial effects that after macromolecular polyvinylpyrrolidone (PVP) is wrapped by the prepared bimetallic nano-enzyme NiCo2O4 with the oxygen vacancy, the stability of the nano-enzyme NiCo2O4 in simulated gastric juice and simulated intestinal juice is remarkably improved. The nano enzyme NiCo2O4-coated PVP can be used for well simulating the enzyme activity of SOD and the enzyme activity of CAT, and can be used for effectively eliminating. OH. The nano-enzyme NiCo2O4-coated PVP is electronegative under neutral and alkaline conditions, and the particle size of the nano-enzyme NiCo2O4-coated PVP is 1t; and the wavelength is 300 nm. The nano-enzyme NiCo2O4-coated PVP has no toxic or side effect, and can effectively eliminate the generation of active oxygen in cells.

The invention relates to the technical field of image description, in particular to a reward in an image description model and a construction method of the image description model, and the construction method of the reward comprises the steps: obtaining label description of a test image and generation description of a preset image description model based on the test image; based on the label description and the generation description, calculating a reward of the n-tuple in the generation description; comparing the size relationship between the reward of the n-tuple and the first threshold value with the size relationship between the reward of each word in the n-tuple and the second threshold value so as to screen out the word with the reward to be increased in the generated description; and according to the label description and the reward of the n-tuple corresponding to the word with the reward to be increased, calculating the reward of the word with the reward to be increased. According to the method, higher rewards are given to words which are not frequently appeared to realize constraint, so that the problem of non-uniform word distribution can be well solved, more specific visual details of an image can be captured, and the description accuracy of the image can be improved.

The invention discloses a facial expression recognition method based on an improved deep convolutional generative adversarial network. The method comprises the steps of 1, acquiring a facial image data set and performing preprocessing; 2, constructing a generative adversarial network, and training the network by using the preprocessed face data set to obtain a composite image; 3, constructing a facial expression classification network, wherein a facial image data set with expression labels is used for training; and 4, connecting the generative adversarial network and the expression classification module to form a whole expression recognition network model, cutting a face image to be recognized into a set size, inputting the face image into the face expression recognition network model, and obtaining an output result to determine a face expression in the face image. The improved generative adversarial network is used for processing the original face image, and finer effective information can be extracted, so that the face expression recognition accuracy is improved.