Patents by Inventor Liang Wei
Liang Wei has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).
-
Publication number: 20240088195Abstract: An image sensor device includes a semiconductor substrate, a radiation sensing member, a shallow trench isolation, and a color filter layer. The radiation sensing member is in the semiconductor substrate. An interface between the radiation sensing member and the semiconductor substrate includes a direct band gap material. The shallow trench isolation is in the semiconductor substrate and surrounds the radiation sensing member. The color filter layer covers the radiation sensing member.Type: ApplicationFiled: November 16, 2023Publication date: March 14, 2024Applicant: TAIWAN SEMICONDUCTOR MANUFACTURING CO., LTD.Inventors: Chia-Yu WEI, Yen-Liang LIN, Kuo-Cheng LEE, Hsun-Ying HUANG, Hsin-Chi CHEN
-
Publication number: 20240087331Abstract: A method for detecting a trailer presence includes identifying a region of interest within a video feed from one or more cameras secured to a vehicle. The video feed depicts a scene having a first pixel area, and the region of interest corresponds to a portion of the scene having a second pixel area that is less than the first pixel area. The method also includes determining an optical flow within the region of interest, and determining that a trailer is connected to the vehicle in response to the optical flow within the region of interest matching a predetermined pattern corresponding to a connected trailer. A camera monitoring system for a vehicle is also disclosed.Type: ApplicationFiled: September 8, 2023Publication date: March 14, 2024Inventors: Wenpeng Wei, Liang Ma, Troy Otis Cooprider
-
Patent number: 11928957Abstract: An audio visual haptic signal reconstruction method includes first utilizing a large-scale audio-visual database stored in a central cloud to learn knowledge, and transferring same to an edge node; then combining, by means of the edge node, a received audio-visual signal with knowledge in the central cloud, and fully mining semantic correlation and consistency between modals; and finally fusing the semantic features of the obtained audio and video signals and inputting the semantic features to a haptic generation network, thereby realizing the reconstruction of the haptic signal. The method effectively solves the problems that the number of audio and video signals of a multi-modal dataset is insufficient, and semantic tags cannot be added to all the audio-visual signals in a training dataset by means of manual annotation. Also, the semantic association between heterogeneous data of different modals are better mined, and the heterogeneity gap between modals are eliminated.Type: GrantFiled: July 1, 2022Date of Patent: March 12, 2024Assignee: NANJING UNIVERSITY OF POSTS AND TELECOMMUNICATIONSInventors: Xin Wei, Liang Zhou, Yingying Shi, Zhe Zhang, Siqi Zhang
-
Publication number: 20240080192Abstract: A data processing method, apparatus and system are disclosed. The method includes: splitting a preset data processing module from a machine learning model, to generate a security application module configured for performing encryption calculation on data input into the machine learning model; taking, through a preset operator, an output value of an operation layer in the split machine learning model as an input value to be input into the security application module, and inputting the input value into the security application module; performing, through the security application module, subgraph calculation in an isolated operating environment according to the input value, to obtain a calculation result; returning the calculation result to the preset operator. The present application solves the technical problem of high computational pressure of TEE model due to that different reasoning frameworks need to be adapted to different customer requirements in the TEE model in related technologies.Type: ApplicationFiled: January 10, 2022Publication date: March 7, 2024Inventors: Caidi WU, Lufei WEI, Liang CHENG
-
Patent number: 11923866Abstract: The present invention discloses a DAC method having signal calibration mechanism is provided. Operation states of current sources are controlled to generate an output analog signal by a DAC circuit according to a codeword of an input digital signal. An echo signal is generated by an echo transmission circuit according to the output analog signal. The codeword is mapped to generate an offset signal by a calibration circuit according to a codeword offset mapping table. The offset signal is processed to generate an echo-canceling signal by an echo-canceling circuit. By a calibration parameter calculation circuit, offset amounts are generated according to a difference between the echo signal and the echo-canceling signal, the offset amounts are grouped to perform statistic operation according to the operation states and current offset values are calculated according to calculation among groups and converted to codeword offset values to update the codeword offset mapping table.Type: GrantFiled: March 1, 2022Date of Patent: March 5, 2024Assignee: REALTEK SEMICONDUCTOR CORPORATIONInventors: Hsuan-Ting Ho, Shih-Hsiung Huang, Liang-Wei Huang
-
Publication number: 20240069299Abstract: An optical element driving mechanism includes a movable assembly, a fixed assembly, and a driving assembly. The movable assembly is configured to be connected to an optical element. The movable assembly is movable relative to the fixed assembly. The driving assembly is configured to drive the movable assembly to move relative to the fixed assembly in a range of motion. The optical element driving mechanism further includes a positioning assembly configured to position the movable assembly at a predetermined position relative to the fixed assembly when the driving assembly is not operating.Type: ApplicationFiled: November 9, 2023Publication date: February 29, 2024Inventors: Chao-Chang HU, Kuen-Wang TSAI, Liang-Ting HO, Chao-Hsi WANG, Chih-Wei WENG, He-Ling CHANG, Che-Wei CHANG, Sheng-Zong CHEN, Ko-Lun CHAO, Min-Hsiu TSAI, Shu-Shan CHEN, Jungsuck RYOO, Mao-Kuo HSU, Guan-Yu SU
-
Publication number: 20240062563Abstract: A system and a method for determining characteristic cells based on image recognition. In the method, a scanning device capturing a full image of the microslide; a host selecting images that comprise stained blocks from the full image; the host sequentially performing image recognition to recognize stained cells of the images, and determining whether the stained cells comprise the characteristic cells with an AI model; and selecting interested images from the images that comprise the characteristic cells, transforming the coordinate system of the interested images into the original coordinate system of the full image, and employing the scanning device to capture the interested images along the Z axis of the original coordinate system of the full image, thereby obtaining and outputting sets of pictures. The present invention can quickly determine whether there are characteristic cells in the tissue under test, so as to provide a diagnostic reference for doctors.Type: ApplicationFiled: December 27, 2022Publication date: February 22, 2024Applicants: V5 TECHNOLOGIES CO., LTD., Taipei Veterans General HospitalInventors: TZU-KUEI SHEN, LINDA SIANA, GUANG-HAO SUEN, LIANG-WEI SHEU, CHIEN-TING YANG, Yuh-Min Chen, Heng-sheng Chao, Chung-Wei Chou, Tsu-Hui Shiao, Yi-Han Hsiao, Chi-Lu Chiang
-
Publication number: 20240048726Abstract: Decoding methods and encoding methods based on an adaptive intra refresh mechanism and related devices are provided. In one aspect, a decoding method includes: receiving a bit stream of a current frame; and determining whether the current frame supports an adaptive intra refresh technology. The determining comprises one of if there is extension data in the bit stream of the current frame and the extension data carries an adaptive intra refresh video extension identifier (ID), obtaining virtual boundary position information carried in the extension data, and determining whether the current frame supports an adaptive intra refresh technology based on the virtual boundary position information; or if there is no adaptive intra refresh video extension ID in the extension data in the bit stream of the current frame, determining that the current frame does not support the adaptive intra refresh technology.Type: ApplicationFiled: March 10, 2022Publication date: February 8, 2024Inventors: Liang WEI, Fangdong CHEN, Li WANG
-
Publication number: 20240047251Abstract: A gas purge device includes a first nozzle and a gas gate. The first nozzle is coupled to a front-opening unified pod (FOUP) through a first port of the FOUP. The gas gate is coupled to the first nozzle via a first pipe. The gas gate includes a first mass flow controller (MFC), a second MFC, and a first switch unit. The first MFC is configured to control a first flow of a first gas. The second MFC is configured to control a second flow of a second gas. The first switch unit is coupled to the first MFC and the second MFC, and is configured to provide the first gas to the first nozzle through the first pipe or receive the second gas from the first nozzle through the first pipe according to a process configuration.Type: ApplicationFiled: October 19, 2023Publication date: February 8, 2024Inventors: MENG-LIANG WEI, SUN-FU CHOU
-
Patent number: 11894856Abstract: The present invention discloses a DAC method having signal calibration mechanism is provided. Operation states of current sources are controlled to generate an output analog signal by a DAC circuit according to a codeword of an input digital signal. An echo signal is generated by an echo transmission circuit according to the output analog signal. The codeword is mapped to generate an offset signal by a calibration circuit according to a codeword offset mapping table. The offset signal is processed to generate an echo-canceling signal by an echo-canceling circuit. By a calibration parameter calculation circuit, offset amounts are generated according to a difference between the echo signal and the echo-canceling signal, the offset amounts are grouped to perform statistic operation according to the operation states and current offset values are calculated according to calculation among groups and converted to codeword offset values to update the codeword offset mapping table.Type: GrantFiled: April 25, 2022Date of Patent: February 6, 2024Assignee: REALTEK SEMICONDUCTOR CORPORATIONInventors: Hsuan-Ting Ho, Liang-Wei Huang, Yun-Chih Tsai, Chia-Lin Chang
-
Publication number: 20240032166Abstract: The present disclosure provides a full-spectrum LED color management system based on RGBWCLA seven-color integration, including an LED lamp; an aluminum substrate inside the LED lamp; a seven-color-integrated full-spectrum integrated LED light source arranged on the aluminum substrate, a lens fixed to the aluminum substrate, a color processing operation chip, a data mining extensions (DMX) communication module, a color control module, a high-precision spectral brightness detector and a color calculation software computer device. In the foregoing mode, the full-spectrum LED color management system implements the following: the seven-color lamp implements a change in simulated blackbody curve correlated color temperature (CCT) from 1,800 K to 8,000 K; the seven-color lamp implements full-spectrum output of white light, and has a color rendering index Ra of 95 or above and R9>90; and the seven-color lamp implements 240% REC.709 color gamut coverage.Type: ApplicationFiled: October 25, 2022Publication date: January 25, 2024Inventors: Yinghui Lu, Liang Wei, Zhangjun Liu
-
Publication number: 20240020994Abstract: A system and a method for cell statistics based on image recognition is disclosed. A microscope captures the full image of a microslide, selects an interest region from the full image, and divides the region into sub-regions. A camera scans and transits the sub-regions to a host. The host performs image recognition on the scanned images to recognize various cells, and accumulates the number of the various cells in each of the scanned images to obtain a cell ratio of each type of the cells. When the number is accumulated and a difference among the cell ratios of the N consecutive scanned images is less than a preset value, the host stops scanning the remains of the sub-regions and outputs the cell ratio of each type of the presently-accumulated cells. The method can obtain various cell ratios of tissue samples for diagnosis.Type: ApplicationFiled: September 23, 2022Publication date: January 18, 2024Applicant: V5 TECHNOLOGIES CO., LTD.Inventors: LINDA SIANA, GUANG-HAO SUEN, LIANG-WEI SHEU, CHIEN-TING YANG
-
Patent number: 11855669Abstract: A method for cancelling radio frequency interference (RFI) and a communication system thereof are provided. In the communication system, digital signals of a frequency domain are converted from analog signals and received by the communication system generally carry RFI, and the signals are processed by an equalizer and a far-end crosstalk canceller. Then, for preventing erroneous signals from forming due to an occurrence of a notch, masking parameters applied to the equalizer and the far-end crosstalk canceller are modified for not processing frequency bands that are RFI-affected. The frequency bands can be ignored by masking corresponding bins in the frequency domain after a fast Fourier transformation. The signals processed by the equalizer and the far-end crosstalk canceller are then outputted to an RFI canceller, and the signals with RFI cancellation can be obtained.Type: GrantFiled: May 31, 2022Date of Patent: December 26, 2023Assignee: REALTEK SEMICONDUCTOR CORP.Inventors: Po-Hsiu Hsiao, Liang-Wei Huang
-
Patent number: 11853890Abstract: Provided is an operation method for a memory device, the memory device being used for implementing an Artificial Neural Network (ANN). The operation method includes: reading from the memory device a weight matrix of a current layer of a plurality of layers of the ANN to extract a plurality of neuro values; determining whether to perform calibration; when it is determined to perform calibration, recalculating and updating a mean value and a variance value of the neuro values; and performing batch normalization based on the mean value and the variance value of the neuro values.Type: GrantFiled: July 26, 2019Date of Patent: December 26, 2023Assignee: MACRONIX INTERNATIONAL CO., LTD.Inventors: Chao-Hung Wang, Yu-Hsuan Lin, Ming-Liang Wei, Dai-Ying Lee
-
Publication number: 20230395586Abstract: A semiconductor structure and processes of forming the same are provided. A semiconductor structure according to the present disclosure includes a first die having a front surface and a back surface and a second die bonded to the back surface of the first die. The first die includes a plurality of trenches adjacent the back surface and the plurality of trenches are filled with a liquid.Type: ApplicationFiled: June 2, 2022Publication date: December 7, 2023Inventors: Tsung-Chieh Hsiao, Chih Hsin Yang, Liang-Wei Wang, Dian-Hau Chen
-
Publication number: 20230395486Abstract: A method of forming semiconductor device includes forming interconnect structure over substrate; forming first passivation layer over the interconnect structure, and metal-insulator-metal capacitor in the first passivation layer; forming first redistribution layer including first pads over the first passivation layer, and first vias extending into the first passivation layer; conformally forming second passivation layer over the first redistribution layer and first passivation layer, and patterning the second passivation layer to form via openings exposing the first pads; forming second redistribution layer including second pads over the second passivation layer, and second vias in the first via openings, wherein the first and second redistribution layers include aluminum-copper alloy and copper, respectively; forming dielectric layer over the second redistribution layer, and patterning the dielectric layer to form via openings exposing some second pads; and forming bumps over the dielectric layer and in theType: ApplicationFiled: June 1, 2022Publication date: December 7, 2023Inventors: Tsung-Chieh HSIAO, Liang-Wei WANG, Dian-Hau CHEN
-
Publication number: 20230389447Abstract: A method of forming a semiconductor device includes providing a bottom electrode; a magnetic tunneling junction (MTJ) element over the bottom electrode; a top electrode over the MTJ element; and a sidewall spacer abutting the MTJ element, wherein at least one of the bottom electrode, the top electrode, and the sidewall spacer includes a magnetic material.Type: ApplicationFiled: August 9, 2023Publication date: November 30, 2023Inventors: Tsung-Chieh Hsiao, Po-Sheng Lu, Wei-Chih Wen, Liang-Wei Wang, Yu-Jen Wang, Dian-Hau Chen, Yen-Ming Chen
-
Publication number: 20230389437Abstract: A semiconductor device includes a substrate with a metal line embedded in the substrate, a dielectric layer disposed on the substrate, a bottom electrode via extending through the dielectric layer and landing on a top surface of the metal line, a bottom electrode disposed on a top surface of the bottom electrode via, a magnetic tunneling junction stack disposed on a top surface of the bottom electrode, and a top electrode disposed on the magnetic tunneling junction stack. A lower portion of the bottom electrode via includes a first metal, and an upper portion of the bottom electrode via includes a second metal that is different from the first metal.Type: ApplicationFiled: July 29, 2023Publication date: November 30, 2023Inventors: Hsiang-Ku Shen, Liang-Wei Wang, Dian-Hau Chen
-
Publication number: 20230352395Abstract: A method for forming a semiconductor structure includes forming a metal-insulator-metal (MIM) structure between first passivation layers over a substrate. The method also includes forming a via structure through the MIM structure and the first passivation layers. The method also includes planarizing the via structure. The method also includes forming an RDL structure over the via structure. The method also includes forming a second passivation layer over the RDL structure and the first passivation layers.Type: ApplicationFiled: April 29, 2022Publication date: November 2, 2023Applicant: Taiwan Semiconductor Manufacturing Company, Ltd.Inventors: Chih-Pin CHIU, Liang-Wei WANG, Chen-Chiu HUANG, Dian-Hau CHEN
-
Publication number: 20230354614Abstract: A semiconductor device includes a substrate, a gate structure, a source region and a drain region, a conductive via and an isolation structure. The gate structure is disposed over the substrate. The source region and the drain region aside the gate structure. The conductive via is disposed in the substrate. The isolation structure is disposed in the substrate, wherein a first surface of the isolation structure is substantially flush with a first surface of the conductive via.Type: ApplicationFiled: July 5, 2023Publication date: November 2, 2023Applicant: Taiwan Semiconductor Manufacturing Company, Ltd.Inventors: Hsiang-Ku Shen, Liang-Wei Wang, Dian-Hau Chen, Yen-Ming Chen