Patents by Inventor Sachin Mehta
Sachin Mehta 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: 20240152973Abstract: The present invention offers systems, methods, and tools for generating real-time coupon/advertisement data-files that may be housed in a database and may be searched for, or retrieved by, data-file users at any time using keyword, subject, and geo-location. Thus, the present invention offers creators of data-files, who may be merchants having businesses, similar features to that of traditional print advertisements but at significantly reduced costs and with much added flexibility. Creators of data-files are able to utilize the systems and methods to create need-specific real-time advertisements, coupons, and offers without online intermediaries or other third party involvement, such as sales persons or marketing teams, resulting in the reduction of overall expenses for marketing the merchant's business.Type: ApplicationFiled: May 7, 2019Publication date: May 9, 2024Inventor: Sachin Mehta
-
Publication number: 20190295260Abstract: A method, a computer readable recording medium, and a system are disclosed for image segmentation using controlled feedback in a neural network. The method includes extracting image data from an image; performing one or more semantic segmentations on the extracted image data; introducing one or more classifiers to each of the one or more semantic segmentations, each of the one or more classifiers assigning a probability to one or more classes of objects within the image; and generating a segmentation mask from the one or more semantic segmentations.Type: ApplicationFiled: October 27, 2017Publication date: September 26, 2019Applicant: Konica Minolta Laboratory U.S.A., Inc.Inventors: Sachin Mehta, Haisong Gu
-
Patent number: 9646358Abstract: A method, device and non-transitory computer readable medium for scene based video watermarking. Methods are disclosed to detect the scenes inside the video. Disclosed devices and methods are capable of scene change detection for both, gradual and abrupt scenes. Methods are also disclosed to group logical scenes in a video. The disclosed methods are aided with methods to reduce the computational time taken for scene change detection. Watermarks are created and segmented using a plurality of unique identifiers to limit any unauthorized use of the video.Type: GrantFiled: September 19, 2014Date of Patent: May 9, 2017Assignee: Infosys LimitedInventors: Sachin Mehta, Rajarathnam Nallusamy
-
Patent number: 9355438Abstract: The geometric distortions of videos and images are corrected wherein a plurality of geometrically distorted frames are mapped with a plurality of original frames of the video content. Further, one or more features associated with the mapped frames are identified as insensitive to the one or more geometric distortions. One or more features of the mapped frames are further mapped with original frames based on a predefined similarity threshold and thereafter one or more geometric distortion parameters are determined. Furthermore, a frame level average distortion and a video level average distortion of each of the one or more geometric distortion parameters are determined, based on which the one or more geometric distortions of the video content are corrected.Type: GrantFiled: January 14, 2015Date of Patent: May 31, 2016Assignee: INFOSYS LIMITEDInventors: Sachin Mehta, Rajarathnam Nallusamy
-
Patent number: 9218638Abstract: The present invention relates to a computer-implemented method, system and computer readable medium for embedding a watermark into a video and extracting a watermark from the original or copy of the watermarked video. The method comprises converting a video frame from RGB color space to YUV color space. Divide a chrominance component into plurality of blocks. Select plurality of blocks based on size of watermark. Associate a pixel in the watermark with selected plurality of blocks. Embed the watermark corresponding to selected plurality of blocks wherein embedding comprises replacing a first pixel value of the block with maximum value of first column of the block if binary value of the corresponding watermark pixel is one or else replace with minimum value. Combine Y component and chrominance components resulting processed YUV video frame and convert it to RGB video frame. Extract the watermark substantially in a reverse process.Type: GrantFiled: June 25, 2012Date of Patent: December 22, 2015Assignee: Infosys LimitedInventors: Sachin Mehta, Vijayaraghavan Varadharajan, Rajarathnam Nallusamy
-
Patent number: 9202257Abstract: A method, non-transitory computer readable medium, and video analyzing computing device that generates a scene averaged frame value for each scene of an original video and a each scene of a resembling video. Each of a subset of the scenes of the resembling video is mapped to a corresponding scene of the original video based on a comparison of the scene averaged frame values. A singular value score is generated for each frame of each of the subset of scenes of the resembling video and each frame of the corresponding one of the scenes of the original video is generated. Matching one(s) of the one or more frames of each of the at least a subset of scenes of the resembling video is identified based on a comparison of at least a subset of the singular value scores. A first watermark is extracted from the identified matching frames.Type: GrantFiled: September 23, 2014Date of Patent: December 1, 2015Assignee: Infosys LimitedInventors: Sachin Mehta, Rajarathnam Nallusamy
-
Publication number: 20150228045Abstract: Method, apparatus and non-transitory computer readable medium for embedding and extracting a watermark in a text document using digital watermarking processes is disclosed. When the text document is watermarked, the following steps are performed. The pages of the text document are transformed into corresponding images. Then, the margins on each of the images are detected and cropped to generate the cropped images. The cropped images are segmented into blocks among which some blocks are selected based on content of each block. The watermark is embedded in the selected blocks using a digital watermarking process. When the watermark from the watermarked text document is extracted, the watermark-embedding process is referred to determine the block information, for selecting each block of a watermarked text document, from where the watermark needs to be extracted.Type: ApplicationFiled: September 23, 2014Publication date: August 13, 2015Inventors: Sachin Mehta, Rajarathnam Nallusamy
-
Publication number: 20150199798Abstract: The technique relates to a system and method for correcting geometric distortions in videos and images. To correct the one or more geometric distortions in videos, the frames of the original and distorted video are mapped at the beginning and then the features associated with the mapped frames which are insensitive to geometric distortions are identified. In case of correcting geometric distortions in images, frame mapping is not required and thus the process starts from identifying the features insensitive to geometric distortions in the original image and distorted image. Then, the geometric distortion parameters are identified from the mapped features. After that, a frame level and video level average distortion of the geometric distortion parameters are determined. Finally, the geometric distortions are corrected based on the frame level and video level average distortion value.Type: ApplicationFiled: January 14, 2015Publication date: July 16, 2015Inventors: Sachin Mehta, Rajarathnam Nallusamy
-
Publication number: 20150170312Abstract: A method, non-transitory computer readable medium, and video analyzing computing device that generates a scene averaged frame value for each scene of an original video and a each scene of a resembling video. Each of a subset of the scenes of the resembling video is mapped to a corresponding scene of the original video based on a comparison of the scene averaged frame values. A singular value score is generated for each frame of each of the subset of scenes of the resembling video and each frame of the corresponding one of the scenes of the original video is generated. Matching one(s) of the one or more frames of each of the at least a subset of scenes of the resembling video is identified based on a comparison of at least a subset of the singular value scores. A first watermark is extracted from the identified matching frames.Type: ApplicationFiled: September 23, 2014Publication date: June 18, 2015Inventors: Sachin Mehta, Rajarathnam Nallusamy
-
Publication number: 20150086067Abstract: A method, device and non-transitory computer readable medium for scene based video watermarking. Methods are disclosed to detect the scenes inside the video. Disclosed devices and methods are capable of scene change detection for both, gradual and abrupt scenes. Methods are also disclosed to group logical scenes in a video. The disclosed methods are aided with methods to reduce the computational time taken for scene change detection. Watermarks are created and segmented using a plurality of unique identifiers to limit any unauthorized use of the video.Type: ApplicationFiled: September 19, 2014Publication date: March 26, 2015Inventors: Sachin Mehta, Rajarathnam Nallusamy
-
Publication number: 20130259294Abstract: The present invention relates to a computer-implemented method, system and computer readable medium for embedding a watermark into a video and extracting a watermark from the original or copy of the watermarked video. The method comprises converting a video frame from RGB color space to YUV color space. Divide a chrominance component into plurality of blocks. Select plurality of blocks based on size of watermark. Associate a pixel in the watermark with selected plurality of blocks. Embed the watermark corresponding to selected plurality of blocks wherein embedding comprises replacing a first pixel value of the block with maximum value of first column of the block if binary value of the corresponding watermark pixel is one or else replace with minimum value. Combine Y component and chrominance components resulting processed YUV video frame and convert it to RGB video frame. Extract the watermark substantially in a reverse process.Type: ApplicationFiled: June 25, 2012Publication date: October 3, 2013Applicant: INFOSYS LIMITEDInventors: Sachin Mehta, Vijayaraghavan Varadharajan, Rajarathnam Nallusamy
-
Publication number: 20130236046Abstract: The present invention relates to a computer-implemented method, system and computer readable medium for detecting source of leakage of a video. The method comprises processing a video by a processing device resulting a processed video, identifying at least one responsible user for safekeeping and/or distributing and/or screening the processed video, creating a watermark wherein the watermark comprises an information about an owner of the video, the responsible user, and at least one transaction information, embedding the watermark inside the processed video resulting a watermarked video, distributing the watermarked video to at least one consumer, and identifying the responsible user by extracting the watermark from a pirated copy of the watermarked video and extracting the information contained in the watermark from the watermarked video.Type: ApplicationFiled: June 20, 2012Publication date: September 12, 2013Applicant: INFOSYS LIMITEDInventors: Rajarathnam Nallusamy, Sachin Mehta