-
公开(公告)号：US11308345B2
公开(公告)日：2022-04-19
申请号：US16848315
申请日：2020-04-14
Applicant: Apple Inc.
Inventor： Vignesh Jagadeesh , Yingjun Bai , Guillaume Tartavel , Gregory Guyomarc'h
Abstract: Various methods for utilizing a saliency heatmaps are described. The methods include obtaining image data corresponding to an image of a scene, obtaining a saliency heatmap for the image of the scene based on a saliency network, wherein the saliency heatmap indicates a likelihood of saliency for a corresponding portion of the scene, and manipulating the image data based on the saliency heatmap. In embodiments, the saliency heatmap may be produced using a trained machine learning model. The saliency heatmap may be used for various image processing tasks, such as determining which portion(s) of a scene to base an image capture device's autofocus, auto exposure, and/or white balance operations upon. According to some embodiments, one or more bounding boxes may be generated based on the saliency heatmap, e.g., using an optimization operation, which bounding box(es) may be used to assist or enhance the performance of various image processing tasks.
-
公开(公告)号：US20210192078A1
公开(公告)日：2021-06-24
申请号：US17129579
申请日：2020-12-21
Applicant: Apple Inc.
Inventor： Stephen Cosman , Kalu Onuka Kalu , Marcelo Lotif Araujo , Michael Chatzidakis , Thi Hai Van Do , Alexis Hugo Louis Durocher , Guillaume Tartavel , Sowmya Gopalan , Vignesh Jagadeesh , Abhishek Bhowmick , John Duchi , Julien Freudiger , Gaurav Kapoor , Ryan M. Rogers
Abstract: Embodiments described herein provide for a non-transitory machine-readable medium storing instructions to cause one or more processors to receive, at a client device, a machine learning model from a server, detect a usage pattern for a content item, store an association between the content item and the detected usage pattern in local data, train the machine learning model using local data for the content item with the detected usage pattern to generate a trained machine learning model, generate an update for the machine learning model, privatize the update for the machine learning model, and transmit the privatized update for the machine learning model to the server.
-
公开(公告)号：US20200380289A1
公开(公告)日：2020-12-03
申请号：US16848315
申请日：2020-04-14
Applicant: Apple Inc.
Inventor： Vignesh Jagadeesh , Yingjun Bai , Guillaume Tartavel , Gregory Guyomarc'h
Abstract: Various methods for utilizing a saliency heatmaps are described. The methods include obtaining image data corresponding to an image of a scene, obtaining a saliency heatmap for the image of the scene based on a saliency network, wherein the saliency heatmap indicates a likelihood of saliency for a corresponding portion of the scene, and manipulating the image data based on the saliency heatmap. In embodiments, the saliency heatmap may be produced using a trained machine learning model. The saliency heatmap may be used for various image processing tasks, such as determining which portion(s) of a scene to base an image capture device's autofocus, auto exposure, and/or white balance operations upon. According to some embodiments, one or more bounding boxes may be generated based on the saliency heatmap, e.g., using an optimization operation, which bounding box(es) may be used to assist or enhance the performance of various image processing tasks.
-
Patent Agency Ranking
公开(公告)号：US20220237887A1
公开(公告)日：2022-07-28
申请号：US17659377
申请日：2022-04-15
Applicant: Apple Inc.
Inventor： Vignesh Jagadeesh , Yingjun Bai , Guillaume Tartavel , Gregory Guyomarc'h
Abstract: Various methods for utilizing a saliency heatmaps are described. The methods include obtaining image data corresponding to an image of a scene, obtaining a saliency heatmap for the image of the scene based on a saliency network, wherein the saliency heatmap indicates a likelihood of saliency for a corresponding portion of the scene, and manipulating the image data based on the saliency heatmap. In embodiments, the saliency heatmap may be produced using a trained machine learning model. The saliency heatmap may be used for various image processing tasks, such as determining which portion(s) of a scene to base an image capture device's autofocus, auto exposure, and/or white balance operations upon. According to some embodiments, one or more bounding boxes may be generated based on the saliency heatmap, e.g., using an optimization operation, which bounding box(es) may be used to assist or enhance the performance of various image processing tasks.
-
公开(公告)号：US20210398333A1
公开(公告)日：2021-12-23
申请号：US16906722
申请日：2020-06-19
Applicant: Apple Inc.
Inventor： Kalu O. Kalu , Guillaume Tartavel
Abstract: Devices, methods, and non-transitory program storage devices are disclosed to provide for automatic cropping of images, given a requested target dimensions and/or aspect ratio, e.g., by using saliency maps to identify the parts of the image containing the most important content—and ensuring that such content is, if possible, included in a determined cropped region from the image. In particular, the devices, methods, and non-transitory program storage devices disclosed herein may: define a first region of interest (ROI) in a given image that is most essential to include in an automatically-determined cropped region; define a second ROI in the given image that would be preferable to include in the automatically-determined cropped region; and then determine a cropped region from the given image, based on the requested target dimensions and/or aspect ratio, that attempts to maximize an amount of overlap between the determined cropped region and the first and/or second ROIs.
-
公开(公告)号：US12189599B2
公开(公告)日：2025-01-07
申请号：US17028920
申请日：2020-09-22
Applicant: Apple Inc.
Inventor： Albert Antony , Francesco Rossi , Guillaume Tartavel , Xiaojin Shi , Marco Zuliani
Abstract: The subject technology provides a framework for evaluating activation functions of a neural network using lookup tables. In order to provide lookup table based activation functions with a desired precision within hardware constraints for the lookup tables, multiple lookup tables for each activation function can be provided. Each of the multiple lookup tables may correspond to a respective subrange of input values, within a full range of input values for the activation function.
-
公开(公告)号：US12052315B2
公开(公告)日：2024-07-30
申请号：US17129579
申请日：2020-12-21
Applicant: Apple Inc.
Inventor： Stephen Cosman , Kalu Onuka Kalu , Marcelo Lotif Araujo , Michael Chatzidakis , Thi Hai Van Do , Alexis Hugo Louis Durocher , Guillaume Tartavel , Sowmya Gopalan , Vignesh Jagadeesh , Abhishek Bhowmick , John Duchi , Julien Freudiger , Gaurav Kapoor , Ryan M. Rogers
IPC: H04L67/1097 , G06F16/2457 , G06F16/438 , G06F16/44 , G06F18/214 , G06F21/62 , G06N3/063 , G06N20/00 , G06V10/774 , G06V10/82 , H04L67/00
CPC classification number: H04L67/1097 , G06F16/24578 , G06F16/438 , G06F16/447 , G06F18/2148 , G06F21/6254 , G06N3/063 , G06N20/00 , G06V10/7747 , G06V10/82 , H04L67/34
Abstract: Embodiments described herein provide for a non-transitory machine-readable medium storing instructions to cause one or more processors to receive, at a client device, a machine learning model from a server, detect a usage pattern for a content item, store an association between the content item and the detected usage pattern in local data, train the machine learning model using local data for the content item with the detected usage pattern to generate a trained machine learning model, generate an update for the machine learning model, privatize the update for the machine learning model, and transmit the privatized update for the machine learning model to the server.
-
公开(公告)号：US11663806B2
公开(公告)日：2023-05-30
申请号：US17659377
申请日：2022-04-15
Applicant: Apple Inc.
Inventor： Vignesh Jagadeesh , Yingjun Bai , Guillaume Tartavel , Gregory Guyomarc'h
IPC: G06F18/214 , G06V10/20 , G06V10/46 , G06V20/64
CPC classification number: G06V10/255 , G06F18/214 , G06V10/462 , G06V20/64
Abstract: Various methods for utilizing a saliency heatmaps are described. The methods include obtaining image data corresponding to an image of a scene, obtaining a saliency heatmap for the image of the scene based on a saliency network, wherein the saliency heatmap indicates a likelihood of saliency for a corresponding portion of the scene, and manipulating the image data based on the saliency heatmap. In embodiments, the saliency heatmap may be produced using a trained machine learning model. The saliency heatmap may be used for various image processing tasks, such as determining which portion(s) of a scene to base an image capture device's autofocus, auto exposure, and/or white balance operations upon. According to some embodiments, one or more bounding boxes may be generated based on the saliency heatmap, e.g., using an optimization operation, which bounding box(es) may be used to assist or enhance the performance of various image processing tasks.
-
-
-
-
-
-
-
Search Results
8
records
(took 0.012 seconds)
