-
公开(公告)号：EP4012546A1
公开(公告)日：2022-06-15
申请号：EP19944564.4
申请日：2019-11-28
发明人： LIU, Haiwei , DONG, Gang , YANG, Hongbin , ZHAO, Yaqian , LI, Rengang , SHI, Hongzhi
摘要： The present invention provides a method and apparatus for data caching. The method comprises: output matrixes are acquired one by one, a plurality of acquired output matrixes are written alternately into two queue sets of a first cache unit according to a sequence in which the output matrixes are acquired, and the output matrixes stored line by line in a first cache unit are written into a second cache unit one by one, according to the sequence in which the output matrixes are written into the second cache unit, valid data of each output matrix of the second cache unit is determined one by one according to preset parameters, and the valid data of each output matrix is written into a third cache unit, and the valid data of the output matrixes stored in the third cache unit are configured to be sequentially written into a memory according to a sequence in which the valid data are written into the third cache unit. In the present solution, the output matrixes are cached by using cache units with the writing speed matching with the computing speed of a processor, and the output matrixes are completely written into a memory one by one according to a sequence of generation time. Therefore, the present invention may solve the problem that the computing speed of the processor does not match with the writing speed of the memory.
-
公开(公告)号：EP4053739A1
公开(公告)日：2022-09-07
申请号：EP20907894.8
申请日：2020-08-25
发明人： WANG, Li , GUO, Zhenhua , ZHAO, Yaqian
摘要： A neural network model for image segmentation, an image segmentation method therefor and device thereof, and a readable storage medium. The model comprises an intelligent selection module, and the intelligent selection module further comprises a feature extraction unit and an intelligent selection unit. Since the feature extraction unit uses multi-scale dilated convolution to obtain information of different scales of an input feature map, a large quantity of diverse feature information is provided for later feature screening. In addition, the intelligent selection unit trains a weight value, and performs intelligent screening on an input feature map channel according to the size of the weight value. Therefore, the intelligent selection module can ensure segmentation accuracy while reducing the number of parameters and the amount of calculation. Therefore, by using the described intelligent selection module, the neural network model of the present application can quickly extract an effective feature of an image; moreover, the amount of calculation is small and model parameters are few, and the model is applicable to a mobile terminal.
-
3.发明公开
公开(公告)号：EP4156079A1
公开(公告)日：2023-03-29
申请号：EP21808636.1
申请日：2021-01-26
发明人： JIANG, Dongdong , ZHAO, Yaqian , DONG, Gang , LI, Rengang , LIU, Haiwei , YANG, Hongbin
摘要： Provided are an image data storage method, an image data processing method and system, and a related apparatus. The image data processing method includes the following steps: sequentially storing image data in a dynamic random memory according to a preset storage format, so that adjacent pieces of image data in the dynamic random memory have continuous storage addresses; reading a preset number of pieces of multi-channel parallel image data from the dynamic random memory, and storing the multi-channel parallel image data in a first-in first-out memory of an FPGA; and executing a convolution operation on target image data in the first-in first-out memory to obtain image feature data. By means of the method, the image data processing rate can be increased.
-
公开(公告)号：EP4012652A1
公开(公告)日：2022-06-15
申请号：EP19944000.9
申请日：2019-12-30
发明人： WANG, Li , GUO, Zhenhua , ZHAO, Yaqian
IPC分类号： G06T7/10
摘要： Provided in the present invention are an image segmentation method and apparatus. A segmented image of each target object, after parameter adjustment, in an acquired image to be segmented is determined by using a preset 3D convolutional neural network model, specifically the process comprising : extracting, by using an extraction module in the 3D convolutional neural network model, a first feature map matrix of at least one target object of the image to be segmented; adjusting, by using a pixel-level significance enhancement module, parameters of the first feature map matrix of each target object, and determining a pixel-level weighting matrix of each target object; enhancing a matrix channel of the first feature map matrix of each target object according to a channel-level significance enhancement module, and determining a channel-level weighting matrix of each target object; performing, by using a 3D residual deconvolutional module, reduction processing on the size of a target matrix of the target object, wherein the size is obtained by increasing the sum of the pixel-level weighting matrix and channel-level weighting matrix of each target object; and determining a segmented image of each target object after parameter adjustment. On the basis of the present invention, a high-precision segmented image can be obtained.
-
-
-
搜索结果
4 条记录 (耗时 0.012 秒)
