8results about How to "Accuracy unchanged" patented technology

Embodiments of the invention provide a small sample target detection method and apparatus, and a computer readable storage medium. The method comprises the following steps of obtaining a to-be-detected picture; performing feature extraction on the to-be-detected picture by adopting an image feature extraction model obtained by pre-training to obtain an actual feature map; analyzing the to-be-detected picture by adopting an object candidate model obtained by pre-training to obtain a plurality of actual detection frames respectively containing corresponding detection objects; and extracting actual detection frame features from the actual feature map by adopting each actual detection frame, and performing similarity matching on each actual detection frame feature and each detection prototype pre-stored in a detection prototype library, and marking the actual detection frame corresponding to the actual detection frame feature with the highest similarity and higher than a preset similarity threshold as a target detection frame, and determining the category of the correspondingly matched detection prototype as the category of the target detection frame. According to the embodiment, the detection efficiency is effectively improved, and the computing power cost is reduced.

This invention discloses a method and apparatus for accelerating distributed training of large AI models based on model prediction. The method includes: constructing a hierarchical parameter dataset; training a prediction model for predicting model parameters of a target model using an evaluation method based on prediction results; computing nodes updating the model parameters of some layers of the target model and then pushing them to a parameter server; on the parameter server, aggregating the received model parameters and predicting the model parameters of the remaining layers of the target model using the prediction model; concatenating the predicted and aggregated model parameters into a complete model parameter set and pushing it to all computing nodes; and replacing the local model parameters with the received complete model parameter set. The prediction model adopts a "convolution + channel attention mechanism + pooling" structure. This invention can significantly reduce communication volume without introducing a large amount of computational overhead while maintaining the accuracy of the target model.

The invention discloses a dual-mode plastic pipe creep ratio performance test device, including a frame, a driving device, a lifting device, a sample positioning device, a loading device, a protective device, an inner diameter measuring device and a control system, wherein the sample positioning device is installed on the bottom plate of the frame, the lifting devices are respectively installed on both sides of the frame, a middle cross beam is arranged on the lifting device, the driving device is arranged at the bottom of the frame and is transmission-connected with the lifting device, a loading device is installed on the middle cross beam, a pressing plate is arranged at the bottom of the loading device, one end of the pressing plate is installed on the middle cross beam through a connecting piece, an inner diameter measuring device is installed on the other end of the pressing plate, and the driving device drives the lifting device to operate and then drives the middle cross beam to move up and down to realize the loading of the sample. The invention has a mechanical and electrical dual-mode loading function, which can be loaded by mechanical weights or by electric control, and solves the problem of the instrument's dependence on the external environment and loading preload during the test.

The present invention discloses a beam assembly of a laser cutting machine in the field of laser cutting machine tools, wherein the beam includes a fixed end and a floating end, a rotating pressure plate 1 is provided on the front side of the floating end, and a rotating pressure plate 2 is provided on the rear side of the floating end; the front and rear sides of the beam corresponding to the floating end are inclined surfaces respectively, the rotating pressure plate 1 and the rotating pressure plate 2 are inclined surfaces corresponding to the beam in the longitudinal direction, and the surfaces of the rotating pressure plate 1 and the rotating pressure plate 2 facing the beam in the horizontal direction are arc-shaped surfaces; the floating end is fixed on the corresponding positioning seat through an elastic component. The present invention ensures that the accuracy of the machine tool will not change due to temperature; when the machine tool fails and the beam collides, it will not cause the straightness of the fuselage guide rail to deviate, the machine tool can recover quickly, and the beam will not be damaged.

The utility model provides an engine connecting rod manufacturing die convenient to demould, and relates to the field of manufacturing dies, the engine connecting rod manufacturing die convenient to demould comprises a processing box, a demoulding assembly is installed on the upper surface of the processing box, two clamping assemblies are installed on the inner wall of the processing box, and a storage assembly is installed on the front face of the processing box. An injection molding assembly is mounted on the upper surface of the machining box. The lower mold in the device is mounted in the rotating demolding assembly, so that after a workpiece is shaped and cooled, the two connecting rods and the rotating plates are driven to rotate through the driving motor, and the lower mold mounted between the two rotating plates can be turned over; by means of the structure, a workpiece in the lower mold can fall down due to the gravity of the workpiece, so that the workpiece can be demolded conveniently, the precision in the lower mold is not changed in the casting or demolding process, and therefore it can be guaranteed that the workpiece is demolded conveniently, and meanwhile the machining quality is guaranteed.

The invention relates to a method for producing a plurality of semiconductor structural elements, in particular power semiconductor structural elements, from a wafer (100), comprising the following steps: providing a wafer substrate (108), treating a structured wafer surface (105), applying an edge insulation (103) to the wafer surface (105), and separating the wafer (100) into individual semiconductor structural elements (102a, 102b, 102c).

According to the data fusion algorithm for the low-altitude radar target and the wireless detection target, the fact that in target information given by a low-altitude radar device and a wireless detection device, only target orientation can enable the low-altitude radar device and the wireless detection device to generate direct relevance is fully utilized, and whether data of the two detection devices are the same target or not is judged through target orientation data. According to the invention, the 0-360-degree detection orientation of the wireless detection device is used as the detection range, after the initial detection orientation is selected, one detection subarea is defined every 15 degrees, and the wireless monitoring target is distributed, so that the characteristic of high target data precision of the low-altitude radar target is fully utilized; the target data of the low-altitude radar target is converted into the target data with the wireless detection device as the center, and the precision of the target data is not changed, so that the difficult problem that whether the target is the same target or not is difficult to judge due to the difference of the target information acquired by the low-altitude radar device and the wireless detection device is solved; and data fusion of a low-altitude radar target and a wireless detection target is realized.

This disclosure relates to a method and electronic device for training an optical proximity correction model. The method includes: selecting a first number of data points as a training subset from a plurality of data points corresponding to measurement points in a layout, using a weighted random strategy, based on multiple index values ​​of the data points; generating updated index values ​​of the data points based on training the model using the training subset; selecting the first number of data points as an updated training subset based on the updated index values ​​of the data points using the weighted random strategy; and performing subsequent iterative training on the model based on the updated training subset. The method of this disclosure can effectively improve the training efficiency of the optical proximity correction model.