Patents Assigned to SENSETIME GROUP LIMITED
-
Patent number: 11900648Abstract: An image generation method includes: extracting a content feature of a first image; respectively extracting an overall image style feature of a second image, and an object style feature of a partial image block, which includes an object, in the second image; determining a target style feature at least according to the overall image style feature and the object style feature; and generating a third image according to the content feature and the target style feature.Type: GrantFiled: October 26, 2021Date of Patent: February 13, 2024Assignees: SENSETIME GROUP LIMITED, HONDA MOTOR CO. LTD.Inventors: Zhiqiang Shen, Mingyang Huang, Jianping Shi, Hideki Matsunaga
-
Publication number: 20240013348Abstract: An image generation method includes the following steps. Illumination lamps on a target object in an original image are detected; the original image is converted into an image to be processed that is in a specific scene; in the image to be processed, at least one illumination lamp of which operation state does not match scene information of the specific scene is determined as an object to be replaced; a replacement image including the object to be replaced of which operation state matches the scene information is determined according to the scene information and the object to be replaced; and the image in a region occupied by the object to be replaced in the image to be processed is replaced with the replacement image to generate a target image.Type: ApplicationFiled: September 22, 2023Publication date: January 11, 2024Applicants: SENSETIME GROUP LIMITED, HONDA MOTOR CO., LTD.Inventors: Guangliang CHENG, Jianping SHI, Yuji YASUI, Hideki MATSUNAGA, Kaname TOMITE
-
Publication number: 20240013453Abstract: An image generation method includes: an original image is converted into an image to be processed under a specific scenario; in the image to be processed, a region to be adjusted of which image rendering state does not match scenario information of the specific scenario is determined; the image rendering state of the region to be adjusted is related to light illuminated on a target object in the original image, and the target object includes a vehicle; a target rendering state in the region to be adjusted is determined; the target rendering state matches the scenario information; and in the image to be processed, the image rendering state of the region to be adjusted is adjusted to the target rendering state, to obtain a target image.Type: ApplicationFiled: September 25, 2023Publication date: January 11, 2024Applicants: SENSETIME GROUP LIMITED, HONDA MOTOR CO., LTD.Inventors: Guangliang CHENG, Jianping SHI, Yuji YASUI, Hideki MATSUNAGA, Kaname TOMITE
-
Patent number: 11710243Abstract: A method for predicting a direction of movement of a target object, a method for training a neural network, a smart vehicle control method, a device, an electronic apparatus, a computer readable storage medium, and a computer program. The method for predicting a direction of movement of a target object comprises: acquiring an apparent orientation of a target object in an image captured by a camera device, and acquiring a relative position relationship of the target object in the image and the camera device in three-dimensional space (S100); and determining, according to the apparent orientation of the target object and the relative position relationship, a direction of movement of the target object relative to a traveling direction of the camera device (S110).Type: GrantFiled: September 18, 2020Date of Patent: July 25, 2023Assignees: SENSETIME GROUP LIMITED, HONDA MOTOR CO. LTD.Inventors: Shu Zhang, Zhaohui Yang, Jiaman Li, Xingyu Zeng
-
Patent number: 11663293Abstract: The disclosure provides an image processing method and device, a computer-readable storage medium. The method includes: performing feature extraction on an image to be processed through a primary image segmentation model to obtain a feature map of the image to be processed; and performing image segmentation on a segmentation target according to the feature map through a target image segmentation model to obtain a target segmentation result, herein the target image segmentation model is obtained from an adversarial training with an image encoding model, and the image encoding model is used for placing a shape constraint on the target segmentation result. The disclosure may improve an accuracy of the image segmentation.Type: GrantFiled: November 30, 2020Date of Patent: May 30, 2023Assignee: SENSETIME GROUP LIMITEDInventors: Yunhe Gao, Rui Huang, Zhiqiang Hu, Hongsheng Li
-
Publication number: 20220044054Abstract: An image generation method includes: extracting a content feature of a first image; respectively extracting an overall image style feature of a second image, and an object style feature of a partial image block, which includes an object, in the second image; determining a target style feature at least according to the overall image style feature and the object style feature; and generating a third image according to the content feature and the target style feature.Type: ApplicationFiled: October 26, 2021Publication date: February 10, 2022Applicants: SENSETIME GROUP LIMITED, HONDA MOTOR CO. LTD.Inventors: Zhiqiang SHEN, Mingyang HUANG, Jianping SHI, Hideki MATSUNAGA
-
Publication number: 20220036587Abstract: A calibration method, an electronic device, and a storage medium are provided. The calibration method includes the following. A sample image is acquired with an image acquisition device. Initial pixel coordinates of multiple sample reference objects in the sample image in an image coordinate system are determined based on the sample image. Straight-line fitting is performed on sample reference objects on a same straight line based on the determined initial pixel coordinates of the sample reference objects in the image coordinate system, and the initial pixel coordinates of the sample reference objects involved in straight-line fitting are corrected based on fitted straight lines to obtain corrected pixel coordinates. A homography matrix of the image acquisition device is determined based on world coordinates of the multiple sample reference objects in the sample image in a world coordinate system and the obtained corrected pixel coordinates.Type: ApplicationFiled: October 15, 2021Publication date: February 3, 2022Applicant: Sensetime Group LimitedInventors: Zheng MA, Rui HUANG, Guohang YAN, Jianping SHI
-
Patent number: 11138756Abstract: A 3D object detection method includes: obtaining two-dimensional (2D) coordinates of at least one predetermined key point of a target object in an image to be processed; constructing a pseudo 3D detection body of the target object in a 2D space according to the 2D coordinates of the at least one predetermined key point; obtaining depth information of a plurality of vertices of the pseudo 3D detection body; and determining a 3D detection body of the target object in a 3D space according to the depth information of the plurality of vertices of the pseudo 3D detection body.Type: GrantFiled: April 1, 2020Date of Patent: October 5, 2021Assignees: SENSETIME GROUP LIMITED, HONDA MOTOR CO. LTD.Inventors: Yingjie Cai, Shinan Liu, Xingyu Zeng, Junjie Yan, Xiaogang Wang, Atsushi Kawamura, Yuji Yasui, Tokitomo Ariyoshi, Yuji Kaneda, Yuhi Goto
-
Publication number: 20210192239Abstract: The present disclosure relates to a method and device for recognizing indication information of indicator lights, an electronic apparatus, and a storage medium. The method comprises: acquiring an input image; determining a detection result of a target object based on the input image, the target object including at least one of an indicator light base and an indicator light in a lighted state, and the detection result including a type of the target object and a position of the target region where the target object in the input image is located; and recognizing, based on the detection result of the target object, the target region where the target object in the input image is located to obtain indication information of the target object.Type: ApplicationFiled: March 5, 2021Publication date: June 24, 2021Applicant: SENSETIME GROUP LIMITEDInventors: Jiabin MA, Zheqi HE, Kun WANG, Xingyu ZENG
-
Publication number: 20210117725Abstract: A method and device for object detection, and a non-transitory computer readable storage medium are provided. The method includes the following. Object detection is performed on images of at least one second domain with a neural network to obtain detection results, where the neural network is trained with a first image sample set for a first domain. For at least one image among the images of the at least one second domain of which the detection result has a confidence level that is lower than a first threshold, the at least one image is assigned as an image sample in at least one second image sample set. At least one image sample is selected from the first image sample set and at least one image sample is selected from each of the at least one second image sample set.Type: ApplicationFiled: December 7, 2020Publication date: April 22, 2021Applicant: Sensetime Group LimitedInventors: Kun WANG, Jiabin MA, Zheqi HE, Hezhang WANG, Xingyu ZENG
-
Publication number: 20210117704Abstract: Implementation modes of the disclosure disclose an obstacle detection method, an intelligent driving control method, an electronic device, and a non-transitory computer-readable storage medium. The obstacle detection method includes that: a first disparity map of an environment image is obtained, the environment image being an image representing information of a space environment where an intelligent device is moving; multiple obstacle pixel areas are determined in the first disparity map; the multiple obstacle pixel areas are clustered to obtain at least one class cluster; and an obstacle detection result is determined according to the obstacle pixel areas belonging to the same class cluster.Type: ApplicationFiled: December 30, 2020Publication date: April 22, 2021Applicant: Sensetime Group LimitedInventors: Xinghua YAO, Xingyu ZHOU, Runtao LIU, Xingyu ZENG
-
Patent number: 10853916Abstract: A method and system for processing an image operates by: filtering a first real image to obtain a first feature map therefor with performances of image features improved; upscaling the obtained first feature map to improve a resolution thereof, the feature map with improved resolution forming a second feature map; and constructing, from the second feature map, a second real image having enhanced performances and a higher resolution than that of the first real image.Type: GrantFiled: June 20, 2018Date of Patent: December 1, 2020Assignee: SENSETIME GROUP LIMITEDInventors: Xiaoou Tang, Chao Dong, Tak Wai Hui, Chen Change Loy
-
Publication number: 20200327690Abstract: A 3D object detection method includes: obtaining two-dimensional (2D) coordinates of at least one predetermined key point of a target object in an image to be processed; constructing a pseudo 3D detection body of the target object in a 2D space according to the 2D coordinates of the at least one predetermined key point; obtaining depth information of a plurality of vertices of the pseudo 3D detection body; and determining a 3D detection body of the target object in a 3D space according to the depth information of the plurality of vertices of the pseudo 3D detection body.Type: ApplicationFiled: April 1, 2020Publication date: October 15, 2020Applicants: SENSETIME GROUP LIMITED, HONDA MOTOR CO. LTD.Inventors: Yingjie CAI, Shinan LIU, Xingyu ZENG, Junjie YAN, Xiaogang WANG, Atsushi Kawamura, Yuji Yasui, Tokitomo Ariyoshi, Yuji Kaneda, Yuhi Goto