45results about How to "Fully explore" patented technology

The invention discloses a multi-dimensional IRT-based teachers teaching ability evaluation method. The evaluation method comprises the following steps: a step of information collection before teachersteaching evaluation, a step of collecting the students evaluation of the teachers classroom language teaching ability which is needed for teachers ability evaluation, a step of preparation for the teachers teaching evaluation, a step of classification analysis on teachers teaching ability, and a step of summarizing the teachers teaching evaluation. The beneficial effects of the method are that via multi-dimensional IRT-based teachers teaching ability evaluation method, comprehensive evaluation of the teachers teaching ability can be realized, operation is simple, the teachers teaching abilityis fully explored, influence exerted by human factors can be prevented, veracity of teachers teaching ability evaluation can be ensured, the method can help reduce the evaluation of teachers teaching, multi-faceted evaluation is adopted, the accuracy of the evaluation of teachers teaching ability can be improved, and evaluation of competent teachers can be facilitated.

The invention relates to a cattle killing machine, and belongs to slaughtering equipment. The cattle killing machine comprises a prisoner tube mechanism, a front push plate, a rear push plate, a cattle rump drive mechanism, a stranglehold mechanism, a cattle head turnover mechanism, a post-slaughter cattle ejecting mechanism, a cavity door start-stop mechanism and a power drive mechanism, wherein the prisoner tube mechanism comprises first and second bases, first and second beams, a top plate, a bottom plate, a left rotary disk, a right rotary disk and a cattle containing groove; one end of the lower part of the front push plate is arranged on the left rotary disk, and the other end of the lower part of the front plate is arranged on the right rotary disk; one end of the lower part of the rear push plate is connected with the left rotary disk, and the other end of the lower part of the rear push plate is connected with the right rotary disk; the cattle rump drive mechanism is arranged on a wall body on one side of the front push plate, and the stranglehold mechanism is arranged on the center position of the first base; the cattle head turnover mechanism is arranged on the stranglehold mechanism, and the post-slaughter cattle ejecting mechanism is arranged on the upper part of one side of the front push plate; the cavity door start-stop mechanism is arranged on the right rotary disk; and the power drive mechanism is arranged on one side of the first base or the second base, the two ends of the first beam are respectively provided with a rear push plate acting cylinder, and the two ends of the second beam are respectively provided with a front push plate action cylinder. The cattle killing machine provided by the invention has a simplified structure, the prisoner tube mechanism is avoided being damaged, and the balance property of a neck clamping force is good.

The invention discloses a light field image super-resolution reconstruction method based on frequency domain analysis and deep learning, which first uses optical flow-based rendering and bicubic interpolation to generate an initial high spatial and angular resolution light field image; then adopts Discrete cosine transform is used to perform frequency domain conversion on the generated initial light field image to obtain a frequency component image that characterizes the characteristics of the light field image; based on this, super-resolution reconstruction can be modeled as a frequency restoration problem, so multiple 2D convolutions are constructed neural network to model the initial restoration of each frequency component, and integrate semantic information into the network to enhance the restoration effect; then combine the frequency components after the initial restoration and build a 3D convolutional neural network for fine restoration; finally, use The inverse discrete cosine transform is used to reconstruct all the restored frequency components into the required light field image; the advantage is that it can effectively improve the spatial and angular resolution of the light field image, and can restore texture information and retain angular consistency.

The invention relates to an environmental detection method and system for archaeology. The environmental detection method comprises the following steps that archaeological excavation is carried out byusing a robot body and a control body; a processor in the robot body or the control body evaluates and divides a passable area and an unpassable area in response to a parallax image collected by a camera device of the robot body; and the processor or the control body releases at least one first-class relay sensor and / or at least one second-class relay sensor based on the unpassable area, and deploys a plurality of third-class relay sensors based on the passable area, wherein the third-class relay sensors are connected mutually in a multi-hop mode and can receive information transmitted by thefirst-class relay sensors and / or the second-class relay sensors.

The present invention provides a deep separable convolutional neural network acceleration method, including: performing deep convolution on the input neuron, and when performing the deep convolution calculation, the same M lines of the C input channel are in the three-dimensional processing unit PE array Independent and parallel computing, to get the output neurons of the same N rows of the C channel, N

The invention belongs to the field of human-machine dialogue, and specifically relates to a knowledge-driven dialogue strategy network optimization method, system, and device, aiming to solve the problem of low utilization rate of training data in the optimization process of the existing dialogue system in the medical field. The method of the present invention is applied to the training of the dialogue system, including: based on the input characteristics of the dialogue samples in the selected field, the self-encoder is optimized by using the collaborative filtering method based on the encoder; the encoder in the self-encoder is used to The input features are mapped to a low-dimensional space; the dialogue policy network is optimized based on a loss function that includes an additional reward value; the additional reward value is inversely proportional to the frequency of state-action pairs in the behavior-action space. The invention can fully explore the state-action space under the condition of limited training data, and improve the data utilization rate.

Belonging to the technical field of biological genetic engineering, the invention in particular relates to a hordeum vulgare calmodulin gene HvCAM1 and salt tolerant application thereof. The hordeum vulgare calmodulin gene HvCAM1 has a nucleotide sequence shown as SEQ ID No:1 and a full-length genome sequence 2124 bp. According to the invention, firstly a hordeum vulgare calmodulin coded gene is obtained by cloning and is named as HvCaM1, and the transcript cDNA full-length nucleotide sequence is shown as SEQ ID No:2, and the full-length cDNA sequence is 728 bp. The invention also utilizes RNAi interference method to silence the HvCAM1 gene successfully in hordeum vulgare Golden Promise, also utilizes the material to verify the negative regulation effect of HvCAM1 gene under salt stress can effectively enhance the salt tolerance of hordeum vulgare. At the same time, overexpression of the HvCAM1 gene in hordeum vulgare Golden Promise cannot enhance the salt tolerance of hordeum vulgare.