公开(公告)号：US20220309764A1

公开(公告)日：2022-09-29

申请号：US17275800

申请日：2020-08-26

Applicant: GOERTEK INC.

Inventor： Jie LIU ,  Jifeng TIAN ,  Fuli XIE ,  Shunran DI ,  Yifan ZHANG

IPC: G06V10/764 ,  G06V10/82 ,  G06V10/26 ,  G06N3/08

Abstract: A method and device for small sample defect classification and a computing equipment are disclosed. The method comprises: separating a target to be tested into parts, and segmenting an original image of the target to be tested into at least two sub-images containing different parts according to the separated parts; establishing small sample classification models with respect to each sub-image and the original image respectively, and obtaining a classification result of each sub-image and a classification result of the original image by using corresponding classification models, wherein the classification result includes a defect category and a corresponding category probability; and determining and outputting a defect category of the target to be tested according to classification results of all the sub-images and the original image.

公开(公告)号：US20210223046A1

公开(公告)日：2021-07-22

申请号：US16631920

申请日：2019-07-31

Applicant: GOERTEK INC.

Inventor： Baoming LI ,  Shanshan MIN ,  Shunran DI ,  Libing ZOU ,  Jinxi CAO

IPC: G01C21/00 ,  G01C21/20 ,  G01S17/86

Abstract: A method and device for extracting key frames in simultaneous localization and mapping and a smart device. The method includes acquiring an image frame from an image library storing a plurality of image frames of an unknown environment, and performing feature extraction on the image frame to obtain information of feature points, wherein the information includes a quantity of feature points; acquiring relative motion information of the image frame relative to the previous key frame, and calculating an adaptive threshold currently used by using the relative motion information; and selecting a key frame according to the information of feature points and the adaptive threshold indicating space information of image frames.

公开(公告)号：US20220309765A1

公开(公告)日：2022-09-29

申请号：US17309308

申请日：2020-10-30

Applicant: GOERTEK INC.

Inventor： Shunran DI ,  Yifan ZHANG ,  Jie LIU ,  Jifeng TIAN

IPC: G06V10/764 ,  G06V10/771 ,  G06V10/776 ,  G06V10/82 ,  G06N3/08

Abstract: The present disclosure discloses a method and device for optimizing an object-class model based on a neural network. The method includes: establishing the object-class model based on the neural network, training the object-class model, and realizing classification of target images by using the object-class model that has been trained; and when a new target image is generated, and the new target image is an image corresponding to a new condition of a target and is capable of still being classified into an original classification system, judging a result of identification of the object-class model to the new target image, and if the object-class model is not capable of correctly classifying the new target image, according to the new target image, selecting some of parameters, adjusting the some of parameters, and training to obtain an object-class model that is capable of correctly classifying the new target image.

公开(公告)号：US20220309640A1

公开(公告)日：2022-09-29

申请号：US17309407

申请日：2020-08-26

Applicant: GOERTEK INC.

Inventor： Jie LIU ,  Jifeng TIAN ,  Fuli XIE ,  Shunran DI ,  Yifan ZHANG

IPC: G06T7/00 ,  G06V10/70 ,  G01N21/88

Abstract: A product defect detection method, device and system are disclosed. The method comprises: acquiring a sample image of a product, extracting candidate image blocks probably including a product defect from the sample image, and extracting preset shape features corresponding to the candidate image blocks and texture features corresponding to the candidate image blocks; training a first-level classifier using the preset shape features to obtain a first-level classifier that can further screen out target image blocks probably including a product defect from the candidate image blocks; training a second-level classifier using the texture features to obtain a second-level classifier that can correctly identify a product defect; and when performing product defect detection, inputting preset shape features of candidate image blocks extracted from a product image into the first-level classifier, and then inputting texture features of obtained target image blocks into the second-level classifier to detect a defect in the product.