-
111.发明公开公开(公告)号：US20230290133A1
公开(公告)日：2023-09-14
申请号：US18016746
申请日：2021-07-27
摘要： Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for selecting actions to be performed by an agent interacting with an environment to accomplish a goal. In one aspect, a method comprises: obtaining an observation characterizing a state of the environment, processing the observation using an embedding model to generate a lower-dimensional embedding of the observation, determining an auxiliary task reward based on a value of a particular dimension of the embedding, determining an overall reward based at least in part on the auxiliary task reward, and determining an update to values of multiple parameters of an action selection neural network based on the overall reward using a reinforcement learning technique.
-
112.发明公开公开(公告)号：US20230290110A1
公开(公告)日：2023-09-14
申请号：US18119152
申请日：2023-03-08
发明人： Robert A. CASE , Joseph JUBINSKI , Dasith A. GUNAWARDHANA , Melvin H. DEDICATORIA , Richard W. HUZIL , Ransom WINDER
CPC分类号： G06V10/70 , G06T7/12 , G06T5/50 , G06T2207/20016 , G06T2207/20132
摘要： Presented herein are systems and methods for generating synthetic training data for machine learning models. Images of a particular object (such as an aircraft) can be received and processed to cutout the object (i.e., separate the object from the background) from the received image. The systems and methods described herein can detect areas in the background images to place an object. Once a suitable area has been detected, the cutout object image can be superimposed on the background image at the location determined to be suitable for placing the object. Superimposing the object onto the background image can include blending the two images using a plurality of blending techniques to reduce artifacts that may bias a supervised training process.
-
公开(公告)号：US20230276115A1
公开(公告)日：2023-08-31
申请号：US17682278
申请日：2022-02-28
CPC分类号： H04N5/23219 , H04N5/247 , G06V40/18 , H04N5/22525 , H04N5/232933 , H04N7/183 , H04N5/23299 , G06V10/70 , G10L15/1815
摘要： An electronic device, a method and a computer program product for selecting an active camera from among a front facing camera and at least one rear facing camera. The method includes capturing, via a front facing camera, an image stream containing a face of a user and determining, via a processor, an eye gaze direction of the user based on an image retrieved from the image stream. The eye gaze direction corresponds to a location where the user is looking. The method further includes in response to determining that the user is looking away from a front surface of an electronic device and towards a direction within a field of view of the at least one rear facing camera, selecting, as an active camera, a corresponding one of the at least one rear facing cameras with a field of view containing the location to which the user is looking.
-
114.发明公开公开(公告)号：US20230267752A1
公开(公告)日：2023-08-24
申请号：US18004175
申请日：2021-07-01
申请人： KYOCERA Corporation
发明人： Takayuki SHINOZAKI
CPC分类号： G06V20/597 , G06V10/70
摘要： An electronic device 10 includes an imaging unit 11, a gaze-line detector 12, and a controller 14. The imaging unit 11 is configured to create at least one image corresponding to a view by image capturing. The gaze-line detector is configured to detect a gaze line of a subject, the gaze line being directed to the view. The controller 14 is configured to serve as a first inference unit 15 and a second inference unit 16. The first inference unit 15 is capable of inferring a gaze-line prediction map based on the at least one image. The second inference unit 16 is capable of inferring biological information regarding an alertness level of the subject based on the gaze-line prediction map and the gaze line of the subject.
-
公开(公告)号：US11734786B2
公开(公告)日：2023-08-22
申请号：US17477282
申请日：2021-09-16
IPC分类号： G06V10/764 , G06T1/20 , G06V10/44 , G06V20/58 , G06V10/70 , G06V30/19 , G06F18/20 , G06V10/82
CPC分类号： G06T1/20 , G06F18/285 , G06V10/454 , G06V10/764 , G06V10/82 , G06V10/87 , G06V20/582 , G06V30/19113 , G06T2207/30252
摘要： Disclosures herein teach applying a set of sections spanning a down-sampled version of an image of a road-scene to a low-fidelity classifier to determine a set of candidate sections for depicting one or more objects in a set of classes. The set of candidate sections of the down-sampled version may be mapped to a set of potential sectors in a high-fidelity version of the image. A high-fidelity classifier may be used to vet the set of potential sectors, determining the presence of one or more objects from the set of classes. The low-fidelity classifier may include a first Convolution Neural Network (CNN) trained on a first training set of down-sampled versions of cropped images of objects in the set of classes. Similarly, the high-fidelity classifier may include a second CNN trained on a second training set of high-fidelity versions of cropped images of objects in the set of classes.
-
公开(公告)号：US11733179B2
公开(公告)日：2023-08-22
申请号：US17085723
申请日：2020-10-30
申请人： ChongQing University
发明人： Yi Qin , Dejun Xi , Caichao Zhu , Sheng Xiang
IPC分类号： G01N21/952 , G06T7/00 , G06V10/22 , G06V10/143 , G06V20/10 , G06V10/70 , G06V10/82 , H04N5/33
CPC分类号： G01N21/952 , G06T7/0004 , G06V10/143 , G06V10/22 , G06V10/70 , G06V10/82 , G06V20/10 , G06T2207/10012 , G06T2207/20081 , G06T2207/30164 , G06V2201/07 , H04N5/33
摘要： The present invention belongs to the field of computer visual detection, and relates to a binocular automatic gear pitting detection device based on deep learning, comprising a gearbox system, a data acquisition system, an image processing system, a tooth surface positioning system, a control system and a motor, wherein the gearbox is used for installing paired meshing gears; the data acquisition system is arranged on the side wall of the gearbox, and a CCD industrial camera is arranged on the data acquisition system; the image processing system completes quantitative evaluation of gear pitting and target detection based on a deep learning technology; both ends of the tooth surface positioning system are respectively connected with the motor and the gearbox, and the torque of the motor is transmitted to an input shaft of the gearbox. The device can determine the optimal installation base points of the data acquisition system according to the characteristics of the meshing gears, and find effective detection areas in combination with the light source and camera arrangement solutions, thereby effectively saving the installation space of the detection device and adapting to the operating characteristics of the meshing gears.
-
公开(公告)号：US20230260297A1
公开(公告)日：2023-08-17
申请号：US18170872
申请日：2023-02-17
IPC分类号： G06V20/69 , G06V10/764 , G06V10/82 , G06V10/70
CPC分类号： G06V20/698 , G06V10/764 , G06V10/82 , G06V10/87 , G06V20/695
摘要： Portable imaging and taxa classification systems and methods are described. One such method comprises accessing at least one microscopic image captured using an imaging device and a microscope from memory and executing a machine learning (ML) routine on the at least one computing device using a single image frame of the at least one microscopic image as an input to identify a presence of a taxa.
-
公开(公告)号：US20230260272A1
公开(公告)日：2023-08-17
申请号：US18172098
申请日：2023-02-21
CPC分类号： G06V10/87 , G06F18/22 , G06T7/001 , G06T2207/20081 , G06T2207/30248
摘要： An image error identification system retrieves an image associated with a vehicle listing and uses various machine learning models to classify the image and generate identification data that may include a vehicle make, model, trim level, and/or various features of the vehicle present in the image. The identification data is compared to the rest of the vehicle listing to detect a mismatch between the image and the vehicle listing. An alert is generated, when a mismatch is detected, indicating the one of the image or the data in the vehicle listing is incorrect.
-
公开(公告)号：US20230260244A1
公开(公告)日：2023-08-17
申请号：US18004769
申请日：2021-05-18
发明人： HAREESH JAGADEESH , KAZUYUKI OKUIKE
CPC分类号： G06V10/147 , G06V10/84 , G06V10/96 , G06V10/955 , H04N25/70 , H04N23/80 , G06V10/87
摘要： For a solid-state imaging element that executes image recognition processing, versatility is improved.
The solid-state imaging element includes a processing section, a digital signal processing section, and an output interface. The processing section selects any one of a plurality of DNNs (Deep Neural Networks) with different formats of an output tensor. The digital signal processing section executes image recognition processing on an input tensor by use of the selected DNN to generate the output tensor. The output interface outputs a decode parameter for decoding the generated output tensor and the output tensor.-
公开(公告)号：US20230259951A1
公开(公告)日：2023-08-17
申请号：US18109204
申请日：2023-02-13
IPC分类号： G06Q30/012 , G06V10/70
CPC分类号： G06Q30/012 , G06V10/70
摘要： A roll-out cart warranty processing system includes a mobile device app and a server. The mobile device app prompts a resident to take at least one image of the roll-out cart, particularly any damaged portion(s) of the roll-out cart. The mobile device app then uploads the at least one image to a server. The server analyzes the at least one image to determine whether the roll-out cart is covered by warranty. The server also analyzes the at least one image to determine whether any identifiable damage would be covered by warranty.
-
-
-
-
-
-
-
-
-
搜索结果
1037 条记录 (耗时 0.033 秒)
