会员体验
专利管家(专利管理)
工作空间(专利管理)
风险监控(情报监控)
数据分析(专利分析)
侵权分析(诉讼无效)
联系我们
交流群
官方交流:
QQ群: 891211   
微信请扫码    >>>
现在联系顾问~
热词
    • 1. 发明申请
    • ELECTRIC VEHICLE
    • WO2022229913A1
    • 2022-11-03
    • PCT/IB2022/053982
    • 2022-04-29
    • POLITECNICO DI TORINO
    • PESCETTO, PaoloPELLEGRINO, Gianmario
    • B60L53/14B60L58/10B60L53/24
    • Electric vehicle (1) comprising at least one recharging apparatus (100) of at least one battery (2), in which said apparatus comprises a power supply station (3) configured to be electrically connected to a source (4) of electric power supply, a conversion station (5), electrically connected to said feeding station (3) and configured to convert said electric feeding power into an electric charging power able to charge said battery (2), said conversion station (5) of said recharging apparatus (100) comprises at least one galvanic separation device (6) electrically interposed between said power station (3) and said battery (2) and configured to electrically separate said battery (2) from said source (4) of electric power supply, in which said galvanic separation device (6) comprises a double winding electric motor (10) suitable for allowing the movement of i said vehicle (1), in which a first winding (11) is electrically connected with said power station (3) and in which a second winding (12) is electrically connected with said battery (2) and magnetically coupled and electrically separated with respect to said first winding (11).
    • 2. 发明申请
    • METHOD FOR DETERMINING THE DEPTH FROM A SINGLE IMAGE AND SYSTEM THEREOF
    • WO2022201212A1
    • 2022-09-29
    • PCT/IT2022/050065
    • 2022-03-22
    • ALMA MATER STUDIORUM - UNIVERSITÀ DI BOLOGNAPOLITECNICO DI TORINO
    • POGGI, MatteoALEOTTI, FilippoTOSI, FabioMATTOCCIA, StefanoPELUSO, ValentinoCIPOLLETTA, AntonioCALIMERA, Andrea
    • G06T7/593
    • Method for determining the depth from a single image and system thereof The present invention relates to a computer-implemented method (2) for determining the depth of a digital image (I), wherein said method (2) comprises the step of training (20) a neural network (3), wherein said training step (20) comprises a first training substep (200) and a second training substep (201), wherein said first training substep (200) comprises the following steps: acquiring (200A) at least one digital image (L, R) of a scene (S), said at least one digital image (L, R) having a first spatial resolution; down-sampling (200B) said at least one digital image (L, R) so as to obtain said digital image (I) being constituted by a matrix of pixels and having a predetermined spatial resolution lower than said first spatial resolution of said at least one digital image (L, R); processing (200C) said digital image (I), obtained in said down-sampling step (200B), through said neural network (3) for generating a first depth map correlated to the depth of each pixel and the surrounding pixels of said digital image (I); and wherein said second training substep (201) comprises the following steps: processing (201A) said at least one digital image (L, R) with a matching technique for generating an optimizing depth map (22) correlated with the depth of each pixel of said at least one digital image (L, R); down-sampling (201B) said optimizing depth map (22) so as to obtain a second depth map with said predetermined spatial resolution lower than said first spatial resolution of said at least one digital image (L, R); determining (201C) a loss function between said first depth map obtained from said first training substep (200) and said second depth map obtained from said second training substep (201), for optimizing said first depth map generated by said neural network (3), wherein the error of said first depth map with respect to said second depth map is used to update the weights of said neural network (3) through back-propagation.