9results about How to "Meet precision requirements" patented technology

The present disclosure relates to a model deployment method and device, electronic equipment, storage medium and product. The method comprises: obtaining an image saliency segmentation model; the image saliency segmentation model is used for saliency segmentation of an input image, the image saliency segmentation model comprises a preset activation function, the output value of the preset activation function is in a preset range, the maximum value of the preset range is a first preset value; converting the model format of the image saliency segmentation model into a target model format supported by a target chip; deploying the image saliency segmentation model in the target model format on the target chip. The present disclosure can limit the size of the output value of the preset activation function, thereby improving the accuracy of the image saliency segmentation model. In addition, by converting the model format of the image saliency segmentation model into the target model format, the image saliency segmentation model can be efficiently run on the target chip, thereby improving the inference efficiency and inference performance of the image saliency segmentation model.

The present application relates to large aperture optical machine structure engineering analysis technical field, especially to a kind of gravity unloading point layout method of large space mirror, comprising the following steps: S1: the finite element analysis model of space mirror is established, and the discrete node surface shape data and unloadable point coordinate data are exported.S2: surface shape fitting is carried out, and surface shape accuracy is calculated.S3: its surface shape residual elevation diagram is drawn, and high residual point is identified.S4: the unloadable point coordinate of space mirror is obtained.S5: space mirror is supported using unloadable point, and its degree of freedom is constrained;Finite element analysis model is updated, and step S2 is repeated, and surface shape accuracy is calculated.S6: whether surface shape accuracy meets the requirements is judged, if not, then repeat step S3 to step S5;Until meet the requirements.The present application has the advantages that, through the gravity unloading point layout of space mirror, gravity unloading point can be more accurately fixed in flexible support mode or the like during processing, and the influence of gravity on surface shape accuracy is reduced.

The application relates to the technical field of CAE preprocessing and cloud cooperation, and discloses a lightweight CAE geometry engine and a cloud real-time cooperation method and system. A cloud server acquires an original CAE geometry model, performs topological simplification and data compression on the original CAE geometry model, and generates a lightweight geometry data file containing multiple resolution levels; the cloud server pushes the received difference data packet to other online cooperation terminals; the other cooperation terminals parse the unique version number carried by the difference data packet through a locally loaded WebAssembly geometry engine module, and perform operation replay updating on the local geometry model according to the differential data contained in the difference data packet. According to the scheme, the original CAE model data volume is reduced by 90%-95%, the WebAssembly engine realizes a 100,000-face model rendering frame rate of 30-60 fps, and the operation response delay is less than or equal to 100 ms, thereby breaking through the performance bottleneck of browser-side geometry processing.

The application provides a SAR image target recognition method based on a saliency map guide, comprising: generating a corresponding target saliency map according to an original SAR image, wherein the target saliency map contains target key information; extracting deep features and shallow features in the original SAR image based on the target saliency map, and guiding the deep features and the shallow features to be fused; refining the fused image features by using a multi-layer hollow convolution to obtain target features; and performing type recognition on the target features by using a full connection layer to obtain a target recognition result. The application avoids noise interference in a background image, improves in-class consistency of target features, can efficiently and robustly realize mining of basic features of a target, and improves precision and efficiency of SAR image target recognition.

The present application belongs to the technical field of molecular medicine, and relates to a probe, a marker combination and a prediction method for non-invasive risk grading of neuroblastoma. The probe UUU-DZ-tFNA can detect UDG in vitro, cells, living bodies and plasma exosomes, and it is proved that UDG in plasma exosomes is closely related to the risk grading of NB. The plasma exosomes of NB patients are collected MYCN Whether to amplify, whether the tumor is metastatic, neuron-specific enolase content, lactate dehydrogenase content and plasma exosome UDG content are integrated, and a NB non-invasive risk grading prediction model is established through a machine learning algorithm. The optimal risk grading model is a neural network model. The area under the receiver operating characteristic curve of the combined model is improved by 5.2% compared with the optimal prediction result of the combined prediction model of the single clinical index, and the sensitivity and specificity are significantly improved. The present application has key clinical values for accurate grading of NB children, individualized treatment plan formulation and efficacy monitoring.

The utility model discloses a kind of coffee bean rationing mechanism, including mechanism shell, welding installation is carried out in the through-hole of mechanism shell upper surface, mechanism shell lower surface is fixedly installed with square funnel in the position of front end, mechanism shell inside lower surface is fixedly installed with electronic metering scale, the surface of electronic metering scale is bonded with coffee bean weighing box, feeding structure is movably installed between mechanism shell inner wall and the lower end of feed pipe, mechanism shell side is welded with integrated controller in the position of a corner, coffee bean weighing box includes C-shaped bottom frame bonded on the surface of electronic metering scale, box body welded on the upper end of C-shaped bottom frame.The utility model said a kind of coffee bean rationing mechanism, by the accurate cooperation of each component, the full-process automation control of the feeding, weighing loading and unloading of coffee bean is realized by this mechanism, satisfy the precision requirement of rationing.

The invention discloses a spin scanning direction finding and cross positioning method based on a single unmanned aerial vehicle platform, and relates to a radio positioning technology. The method comprises the following steps: controlling an unmanned aerial vehicle carrying a directional antenna to hover at a first measurement point and spin around an axis, and synchronously acquiring received signal strength (RSSI) and a yaw angle; obtaining a first azimuth angle of the radiation source by detecting an RSSI peak value; then the unmanned aerial vehicle flies to a second measurement point to repeat the process to obtain a second azimuth angle; and finally, calculating the coordinates of the radiation source by using a cross positioning principle according to the coordinates and the azimuth angles of the measurement points. According to the invention, the antenna mechanical scanning is realized by using the spinning of the unmanned aerial vehicle, the single-station multi-point direction finding and positioning are realized by using extremely simple hardware (single antenna), and the method has the advantages of low cost, fast deployment and easy integration, and is suitable for fast searching and positioning of a ground fixed radiation source.

The invention discloses a lightweight positioning method and system based on two-dimensional laser relative cubessatellites, and the method comprises the steps: collecting two-dimensional laser radar polar coordinate point cloud data of the cubessatellites, converting the data into a two-dimensional raster image, carrying out the preprocessing, obtaining a denoised two-dimensional raster image, and carrying out the positioning of the cubessatellites based on the image. A lightweight YOLO model is adopted to select an initial target point cloud cluster region; the method comprises the following steps: acquiring contour feature parameters of an initial target point cloud cluster region, obtaining a three-dimensional relative position observation vector of a cubesat relative to the center of mass of the aircraft, obtaining a two-dimensional pose of the aircraft in a global coordinate system by using a multi-layer perceptron neural network based on the three-dimensional relative position observation vector and historical pose information, and completing positioning. According to the method, stable and accurate relative pose estimation can still be realized under the conditions of noise interference and target dynamic change, and a reliable visual positioning solution is provided for a lightweight satellite autonomous docking task.

The utility model relates to a vehicle-mounted weighing device, which comprises a pressure data acquirer, a vehicle-mounted weighing controller and a display, the pressure data acquirer comprises a plurality of suspension embedded pressure sensors, the suspension embedded pressure sensors are uniformly arranged on a vehicle suspension, and the suspension embedded pressure sensors are electrically connected with the vehicle-mounted weighing controller through cables with shielding layers; the vehicle-mounted weighing controller is fixed to the inner wall of a vehicle cab, the displayer is embedded into a center console of a vehicle, and the vehicle-mounted weighing controller is connected with the displayer through a CAN bus interface. The pressure data acquisition device acquires pressure data borne by the current vehicle and sends the pressure data to the vehicle-mounted weighing controller; and the vehicle-mounted weighing controller reads the pressure data, sends the pressure data to an external data platform by using an internal communicator, and sends the returned load data to a display for displaying. The device can break through technical bottlenecks among data accuracy, operation and maintenance efficiency and working condition adaptability.