公开(公告)号：US11961277B2

公开(公告)日：2024-04-16

申请号：US17562980

申请日：2021-12-27

Applicant: Tsinghua University

Inventor： Gao Huang ,  Shiji Song ,  Haojun Jiang ,  Le Yang ,  Yiming Chen

IPC: G06V10/00 ,  G06F18/22 ,  G06V10/44 ,  G06V10/77 ,  G06V10/774

CPC classification number: G06V10/443 ,  G06F18/22 ,  G06V10/7715 ,  G06V10/774

Abstract: A method for detecting image information includes: acquiring at least one sample of image pair to be processed; calculating a reconstruction loss function of the second feature extraction model based on the first image samples and the first reconstructed image feature information; calculating an adversarial loss function of the third feature extraction model based on the second reconstructed image feature information and the first image samples; optimizing the first model parameters in the first feature extraction model based on the reconstruction and the adversarial loss function to generate the optimized first feature extraction model; inputting the acquired image pair to be processed into the optimized first feature extraction model to generate the difference information. The method reduces the first feature extraction model's dependence on the labeled data and improves the model's recognition efficiency and accuracy by using the samples without the labeled difference information.

公开(公告)号：US12033365B2

公开(公告)日：2024-07-09

申请号：US17562945

申请日：2021-12-27

Applicant: Tsinghua University

Inventor： Gao Huang ,  Shiji Song ,  Chaoqun Du

IPC: G06V10/26 ,  G06T7/11 ,  G06V10/77 ,  G06V10/774 ,  G06V10/776 ,  G16H30/40 ,  G16H50/20 ,  G06V10/82

CPC classification number: G06V10/26 ,  G06T7/11 ,  G06V10/7715 ,  G06V10/7747 ,  G06V10/776 ,  G16H30/40 ,  G16H50/20 ,  G06T2207/10081 ,  G06V10/82 ,  G06V2201/031

Abstract: An image processing method and apparatus and a storage medium, wherein the method particularly includes firstly acquiring an image-to-be-trained sample and a label segmentation image corresponding to the image-to-be-trained sample; inputting the image-to-be-trained sample into an image segmentation model to be trained, obtaining a first image feature of a last one output layer in the image segmentation model and a second image feature of a second last output layer when the image-to-be-trained sample is being extracted by using the image segmentation model, outputting the corresponding segmented-image samples; based on the label segmentation image and the segmented-image samples, calculating the model loss function, optimizing the model parameter, and generating the image segmentation model that has been optimized; and inputting an acquired image to be processed into the image segmentation model that has been optimized, and generating segmented images corresponding to the image to be processed.

公开(公告)号：US20240202497A1

公开(公告)日：2024-06-20

申请号：US18572377

申请日：2021-07-21

Applicant: ROBERT BOSCH GMBH ,  TSINGHUA UNIVERSITY

Inventor： Chunjiang Ge ,  Gao Huang ,  Rui Lu ,  Shiji Song ,  Xuran Pan ,  Hao Yang

IPC: G06N3/045 ,  G06N3/0464

CPC classification number: G06N3/045 ,  G06N3/0464

Abstract: A method for computer vision processing. The method includes projecting input visual data into a plurality of intermediate feature maps by performing a plurality of 1×1 convolution operations; generating an attention weighted map by performing attention and aggregation operations on the plurality of intermediate feature maps; generating a convolved feature map by performing shift and summation operations on the plurality of intermediate feature maps; and adding the attention weighted map and the convolved feature map based on at least one scalar.

公开(公告)号：US11638952B2

公开(公告)日：2023-05-02

申请号：US17708000

申请日：2022-03-30

Applicant: Tsinghua University

Inventor： Shiji Song ,  Shengsheng Niu ,  Yali Chen

IPC: B22D11/16 ,  B22D11/18

Abstract: A steelmaking-and-continuous-casting dispatching method and apparatus based on a distributed robust chance-constraint model. The method includes: according to parameters, an objective function and a constraint condition in steelmaking-and-continuous-casting dispatching, establishing the distributed robust chance-constraint model; by using a dual-approximation method or a linear-programming-approximation method, solving the distributed robust chance-constraint model, to obtain processing starting durations of cast batches in conticasters and processing starting durations of furnace batches in machines other than the conticasters; and by using a solved result of the distributed robust chance-constraint model as an evaluation criterion, by using a tabu-search algorithm, determining a furnace-batch sequence and a distribution theme in the steelmaking-and-continuous-casting dispatching. The method deems the processing duration in the steelmaking-and-continuous-casting process as a random variable, and makes the description by using the polyhedral support set and the accurate moment information, and the method meets the actual production conditions more than the conventional research models.