-
公开(公告)号:WO2023091925A1
公开(公告)日:2023-05-25
申请号:PCT/US2022/079927
申请日:2022-11-16
Applicant: QUALCOMM INCORPORATED
Inventor: BORSE, Shubhankar Mangesh , PARK, Hyojin , CAI, Hong , DAS, Debasmit , GARREPALLI, Risheek , PORIKLI, Fatih Murat
IPC: G06V10/20 , G06V10/764 , G06V20/70
Abstract: Aspects of the present disclosure relate to a novel framework for integrating both semantic and instance contexts for panoptic segmentation. In one example aspect, a method for processing image data includes: processing semantic feature data and instance feature data with a panoptic encoding generator to generate a panoptic encoding; processing the panoptic encoding to generate a panoptic segmentation features; and generating the panoptic segmentation mask based on the panoptic segmentation features.
-
公开(公告)号:WO2023091444A1
公开(公告)日:2023-05-25
申请号:PCT/US2022/050037
申请日:2022-11-16
Applicant: SIMPLISAFE, INC.
Inventor: JACOB, Jacquilene
Abstract: The techniques described herein relate to computerized methods, systems and non-transitory computer-readable media for determining a plurality of regions of interest from an image of a scene for motion detection. The methods can include generating the regions of interest using image segmentation techniques and receiving user selection to designate one or more regions as motion detection zones. The methods can also automatically recommend motion detection zones. The methods can include subsequently capturing one or more images of a scene and performing motion detection in the one or more images of the scene using the designated motion detection zones.
-
公开(公告)号:WO2023091211A1
公开(公告)日:2023-05-25
申请号:PCT/US2022/042284
申请日:2022-09-01
Applicant: MICROSOFT TECHNOLOGY LICENSING, LLC.
Inventor: TIWARI, Nidhi , HAN, Chao , GURDASANI, Krish
Abstract: Methods, systems, and computer storage media for providing a tailored meeting-video segment associated with a meeting-video management engine of a meeting-video management system. The tailored meeting-video segment corresponds to a portion of meeting-video content that is programmatically generated based on features associated with video data, meeting data, and user data. A tailored meeting-video segment – or a plurality of tailored meeting-video segments – can be generated by employing a meeting-video tailoring machine learning model of the meeting-video management engine. In particular, the features – associated with video data comprising the plurality of clips, meeting data of the meeting, and user data of the user – are meeting-video tailoring features used by the meeting-video tailoring machine learning model to generate the tailored meeting-video segment. The tailored meeting-video segment is communicated to a user to enable uniquely tailored presentation and playback of meeting-video content computed to be relevant to the user via the meeting-video management engine.
-
公开(公告)号:WO2023089637A1
公开(公告)日:2023-05-25
申请号:PCT/IN2022/051015
申请日:2022-11-19
Applicant: FLYING FLAMINGOS INDIA PVT. LTD.
Inventor: GUPTA, Rajat , AGARWAL, Shourya , PATIL, Malhar , SRIVASTAVA, Prakharkumar , KUSHWAHA, Kalpit Singh
Abstract: A system for presenting hyper-personalized content over objects is provided. The system determines context features based on digital media, sensor data, and a user profile of a user of a mobile device. The system computes a context vector representing a context associated with the digital media and the user based on a correlation of the context features and compares the context vector with a plurality of context vectors of a plurality of multimedia content. The system selects, from the plurality of multimedia content, a multimedia content based on the comparison of the context vector with the plurality of context vectors. The system renders, on a display of the mobile device, an augmented reality presentation in which the selected multimedia content is superimposed on a target object displayed in the digital media. The augmented reality presentation is hyper-personalized to map to the context associated with the user.
-
公开(公告)号:WO2023088438A1
公开(公告)日:2023-05-25
申请号:PCT/CN2022/132926
申请日:2022-11-18
Applicant: 维沃移动通信有限公司
Inventor: 单昌鹏
IPC: G06V40/13
Abstract: 一种屏幕模组、电子设备和屏幕模组的制造方法。屏幕模组包括显示模组、保护板和指纹模组;保护板设于显示模组上;指纹模组设于保护板下表面,位于显示模组与保护板之间。屏幕模组将指纹模组设置于保护板下表面,使得指纹模组处于显示模组的上方,在用户使用指纹模组时,指纹模组与人体的手指配合,进而实现对指纹的采集。
-
公开(公告)号:WO2023088314A1
公开(公告)日:2023-05-25
申请号:PCT/CN2022/132308
申请日:2022-11-16
Applicant: 王树松
Inventor: 王树松 WANG, Shusong
IPC: G06V10/00
Abstract: 一种对象分类方法、装置、设备及存储介质。该方法包括:获取待分类对象;基于对象分类网络对待分类对象进行分类,得到待分类对象的分类结果,对象分类网络中的至少一个目标隐藏层的神经元基于抑制修正线性单元进行激活;抑制修正线性单元包括线性抑制函数和修正线性类函数,修正线性类函数使用线性抑制函数的输出作为输入;线性抑制函数由输入与对应的线性抑制系数相乘的乘积组成;修正线性类函数为正向修正线性函数或者负向修正线性函数;线性抑制系数为小于1的正数值,且目标神经元的线性抑制系数的值与目标神经元所属目标隐藏层其他神经元的对应线性抑制系数的值相同。由此,引入抑制修正线性单元的对象分类网络提高了分类准确性。
-
公开(公告)号:WO2023088087A1
公开(公告)日:2023-05-25
申请号:PCT/CN2022/129026
申请日:2022-11-01
Applicant: 华为云计算技术有限公司 , 北京构力科技有限公司
IPC: G06V30/422 , G06K9/62 , G06V30/19
Abstract: 本申请提供一种设计图转换方法、装置及相关设备,用于对设计图进行转换以得到三维模型,该方法包括:首先获取目标物的平面图和第一设计图,该第一设计图是目标物的剖面图和/或立面图;然后通过人工智能模型识别平面图和第一设计图中的各个构件和各组标注字符,输出识别结果,该识别结果包括各个构件的类型、位置与范围,以及各组标注字符的内容与位置;再根据识别结果将目标物的二维的设计图转换为三维模型。通过人工智能模型对一个目标物的多种类型的设计图中的构件和标注字符进行识别,根据识别结果得到该目标物的设计数据,进而根据设计数据对目标物进行转换得到对应的三维模型,能够提高转换的效率和准确性。
-
公开(公告)号:WO2023087659A1
公开(公告)日:2023-05-25
申请号:PCT/CN2022/095363
申请日:2022-05-26
Applicant: 浪潮(北京)电子信息产业有限公司
Abstract: 本申请公开了一种多模态数据处理方法、装置、设备及存储介质,该方法包括:获取目标物体的不同光学模态信息,制作多模态数据集;构建多模态融合网络模型;多模态融合网络模型包括用于提取各模态特征的模态特征提取网络,用于将各模态特征进行合并的模态特征融合网络,以及用于将合并后的目标特征进行分类任务或回归任务的决策网络;利用多模态数据集训练多模态融合网络模型;获取待测物体的不同光学模态信息,并输入至训练完成的多模态融合网络模型中,输出分类结果或回归结果。
-
公开(公告)号:WO2023087597A1
公开(公告)日:2023-05-25
申请号:PCT/CN2022/083740
申请日:2022-03-29
Applicant: 苏州浪潮智能科技有限公司
Abstract: 一种图像处理方法、系统、计算机设备以及可读存储介质,所述方法包括以下步骤:对初始数据集中的图像进行预处理以得到训练数据集(S1);利用训练数据集对图像分割神经网络进行训练(S2);将训练后的图像分割神经网络的最后一层损失函数层去除后得到推理网络(S3);将训练数据集输入到推理网络中以得到多个逻辑向量(S4);根据多个逻辑向量、初始数据集以及初始数据集中的每一个图像的掩膜对校验网络进行训练(S5);利用推理网络和训练好的校验网络对待处理图像进行推理以得到待处理图像的掩膜(S6)。该方法针对高分辨率图像在大规模图像分割网络训练时内存溢出的情况提出了一种解决方案,保障图像分割精度的同时,降低网络训练所需显存。
-
公开(公告)号:WO2023087420A1
公开(公告)日:2023-05-25
申请号:PCT/CN2021/135634
申请日:2021-12-06
Applicant: 南京航空航天大学
IPC: G06V40/20
Abstract: 本发明公开一种基于热红外视觉的停机坪人体动作识别方法及系统,该方法包括:从红外监控视频中获取多个视频序列;对视频序列中每帧图像中的设定目标进行目标框标注;对于视频序列中每帧图像,根据标注后的目标框截取目标框放大区域;将目标框标注图像的位置信息添加到目标框放大区域,获得三通道子图像;各三通道子图像按时间顺序构成三通道子图像序列;将多个视频序列对应的三通道子图像序列作为训练集对动作识别模型进行训练;从红外监控视频中获取待识别视频序列,获得待识别视频序列对应的三通道子图像序列;将待识别视频序列对应的三通道子图像序列输入训练好的动作识别模型输出目标动作类型。本发明提高了复杂环境下人体动作的识别精度。
-
-
-
-
-
-
-
-
-