Patents by Inventor Mithun Das Gupta
Mithun Das Gupta 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).
-
Patent number: 11182408Abstract: A computer-implemented technique is described herein for using a machine-trained model to identify individual objects within images. The technique then creates a relational index for the identified objects. That is, each index entry in the relational index is associated with a given object, and includes a set of attributes pertaining to the given object. One such attribute identifies at least one latent semantic vector associated with the given object. Each attribute provides a way of linking the given object to one or more other objects in the relational index. In one application of this technique, a user may submit a query that specifies a query object. The technique consults the relational index to find one or more objects that are related to the query object. In some cases, the query object and each of the other objects have a complementary relationship.Type: GrantFiled: May 21, 2019Date of Patent: November 23, 2021Assignee: Microsoft Technology Licensing, LLCInventors: Kun Wu, Yiran Shen, Houdong Hu, Soudamini Sreepada, Arun Sacheti, Mithun Das Gupta, Rushabh Rajesh Gandhi, Sudhir Kumar
-
Patent number: 10949706Abstract: A computer-implemented technique is described herein for retrieving at least one recommended output image. In one implementation, the technique uses a generator component to transform first-part image information, associated with a first-part image selected by a user, into one or more instances of second-part generated image information. Each instance of the second-part generated image information complements the first-part image information. The generator component is trained by a computer-implemented training system using a conditional generative adversarial network (cGAN). The technique further includes: retrieving one or more second-part output images from a data store based on the instance(s) of second-part generated image information; generating a user interface presentation that presents the first-part image and the second-part output image(s); and displaying the user interface presentation on a display device.Type: GrantFiled: January 16, 2019Date of Patent: March 16, 2021Assignee: Microsoft Technology Licensing, LLCInventors: Mithun Das Gupta, Sudhir Kumar, Rushabh Rajesh Gandhi, Soudamini Sreepada, Naresh Annam, Shashank Verma, Jagjot Singh
-
Publication number: 20200372047Abstract: A computer-implemented technique is described herein for using a machine-trained model to identify individual objects within images. The technique then creates a relational index for the identified objects. That is, each index entry in the relational index is associated with a given object, and includes a set of attributes pertaining to the given object. One such attribute identifies at least one latent semantic vector associated with the given object. Each attribute provides a way of linking the given object to one or more other objects in the relational index. In one application of this technique, a user may submit a query that specifies a query object. The technique consults the relational index to find one or more objects that are related to the query object. In some cases, the query object and each of the other objects have a complementary relationship.Type: ApplicationFiled: May 21, 2019Publication date: November 26, 2020Inventors: Kun WU, Yiran SHEN, Houdong HU, Soudamini SREEPADA, Arun SACHETI, Mithun Das GUPTA, Rushabh Rajesh GANDHI, Sudhir KUMAR
-
Patent number: 10740940Abstract: Techniques for automating the generation and analysis of fundus drawings are described. Captured images undergo image processing to extract information about image features. Fundus images are generated and recommended labels for the fundus drawing are generated. Fundus drawings can be analyzed and undergo textual processing to extract existing labels. Machine learning models and co-occurrence analysis can be applied to collections of fundus images and drawings to gather information about commonly associated labels, label locations, and user information. The most frequently used labels associated with the image can be identified to improve recommendations and personalize labels.Type: GrantFiled: November 26, 2018Date of Patent: August 11, 2020Assignee: Microsoft Technology Licensing, LLCInventors: Prashant Gupta, Manish Gupta, Mithun Das Gupta
-
Publication number: 20200226411Abstract: A computer-implemented technique is described herein for retrieving at least one recommended output image. In one implementation, the technique uses a generator component to transform first-part image information, associated with a first-part image selected by a user, into one or more instances of second-part generated image information. Each instance of the second-part generated image information complements the first-part image information. The generator component is trained by a computer-implemented training system using a conditional generative adversarial network (cGAN). The technique further includes: retrieving one or more second-part output images from a data store based on the instance(s) of second-part generated image information; generating a user interface presentation that presents the first-part image and the second-part output image(s); and displaying the user interface presentation on a display device.Type: ApplicationFiled: January 16, 2019Publication date: July 16, 2020Inventors: Mithun Das GUPTA, Sudhir KUMAR, Rushabh Rajesh GANDHI, Soudamini SREEPADA, Naresh ANNAM, Shashank VERMA, Jagjot SINGH
-
Patent number: 10575789Abstract: A method and apparatus is disclosed herein for erythema grading for psoriasis. In one embodiment, the method comprises generating Skellam distribution statistics related to differences between readout voltages for one or more psoriasis regions for a patient and a normal skin region for the patient; and generating an erythema grading classification for erythema regions in psoriasis using the Skellam distribution statistics as feature vectors.Type: GrantFiled: December 5, 2014Date of Patent: March 3, 2020Assignee: RICOH CO., LTD.Inventor: Mithun Das Gupta
-
Patent number: 10354754Abstract: Embodiments digitize radiology films into DICOM format. Radiology films typically include an array of captured images laid out in a grid pattern. To comply with DICOM format, the scanned image of a radiology film is segmented into sub-images and text is extracted from the sub-images to generate DICOM metadata. The sub-images and extracted text metadata are then combined to generate a DICOM-compliant multi-image file.Type: GrantFiled: October 16, 2017Date of Patent: July 16, 2019Assignee: RICOH CO., LTD.Inventors: Pritesh Sankhe, Mithun Das Gupta
-
Publication number: 20190096111Abstract: Techniques for automating the generation and analysis of fundus drawings are described. Captured images undergo image processing to extract information about image features. Fundus images are generated and recommended labels for the fundus drawing are generated. Fundus drawings can be analyzed and undergo textual processing to extract existing labels. Machine learning models and co-occurrence analysis can be applied to collections of fundus images and drawings to gather information about commonly associated labels, label locations, and user information. The most frequently used labels associated with the image can be identified to improve recommendations and personalize labels.Type: ApplicationFiled: November 26, 2018Publication date: March 28, 2019Inventors: Prashant Gupta, Manish Gupta, Mithun Das Gupta
-
Patent number: 10163241Abstract: Techniques for automating the generation and analysis of fundus drawings are described. Captured images undergo image processing to extract information about image features. Fundus images are generated and recommended labels for the fundus drawing are generated. Fundus drawings can be analyzed and undergo textual processing to extract existing labels. Machine learning models and co-occurrence analysis can be applied to collections of fundus images and drawings to gather information about commonly associated labels, label locations, and user information. The most frequently used labels associated with the image can be identified to improve recommendations and personalize labels.Type: GrantFiled: December 9, 2016Date of Patent: December 25, 2018Assignee: Microsoft Technology Licensing, LLCInventors: Prashant Gupta, Manish Gupta, Mithun Das Gupta
-
Publication number: 20180165850Abstract: Techniques for automating the generation and analysis of fundus drawings are described. Captured images undergo image processing to extract information about image features. Fundus images are generated and recommended labels for the fundus drawing are generated. Fundus drawings can be analyzed and undergo textual processing to extract existing labels. Machine learning models and co-occurrence analysis can be applied to collections of fundus images and drawings to gather information about commonly associated labels, label locations, and user information. The most frequently used labels associated with the image can be identified to improve recommendations and personalize labels.Type: ApplicationFiled: December 9, 2016Publication date: June 14, 2018Inventors: Prashant Gupta, Manish Gupta, Mithun Das Gupta
-
Patent number: 9965588Abstract: Embodiments digitize radiology films into DICOM format. Radiology films typically include an array of captured images laid out in a grid pattern. To comply with DICOM format, the scanned image of a radiology film is segmented into sub-images and text is extracted from the sub-images to generate DICOM metadata. The sub-images and extracted text metadata are then combined to generate a DICOM-compliant multi-image file.Type: GrantFiled: March 6, 2014Date of Patent: May 8, 2018Assignee: RICOH CO., LTD.Inventors: Pritesh Sankhe, Mithun Das Gupta
-
Publication number: 20180075191Abstract: Embodiments digitize radiology films into DICOM format. Radiology films typically include an array of captured images laid out in a grid pattern. To comply with DICOM format, the scanned image of a radiology film is segmented into sub-images and text is extracted from the sub-images to generate DICOM metadata. The sub-images and extracted text metadata are then combined to generate a DICOM-compliant multi-image file.Type: ApplicationFiled: October 16, 2017Publication date: March 15, 2018Applicant: Ricoh Company, Ltd.Inventors: Pritesh Sankhe, Mithun Das Gupta
-
Patent number: 9675247Abstract: Retinal vessel extraction on color images of eyes are disclosed. In one embodiment, the method comprises performing alpha matting on a multi-dimensional feature set derived from image data of a first image and performing retinal vessel extraction, including performing segmentation on the image by separating foreground and background image data from the first image using an output from the alpha matting.Type: GrantFiled: December 5, 2014Date of Patent: June 13, 2017Assignee: RICOH CO., LTD.Inventors: Powel Talwar, Mithun Das Gupta
-
Patent number: 9582890Abstract: Embodiments segment received image data. The received image data may include pixels that have multiple channels of intensity values. The image data is decomposed into albedo and shading components. This may be accomplished using a minimization that enforces a relationship between albedo, shading, and intensity values. The minimization may also include an albedo regularizer to infer albedo in part based on chromaticity and albedo of surrounding pixels. Superpixels are generated based on contiguous regions of pixels having similar image data across channels. These superpixels are then merged based in part on the determined albedo and shading components as well as based on the image data. The channels of image data may include infrared image data used to modify visible channels of the image to create pseudo-image data, which may be used in place of image data for albedo-shading decomposition, superpixel generation, or superpixel merging.Type: GrantFiled: September 19, 2014Date of Patent: February 28, 2017Assignee: Ricoh Company, LTD.Inventors: Mithun Das Gupta, Srinidhi Srinivasa
-
Publication number: 20160157786Abstract: A method and apparatus is disclosed herein for erythema grading for psoriasis. In one embodiment, the method comprises generating Skellam distribution statistics related to differences between readout voltages for one or more psoriasis regions for a patient and a normal skin region for the patient; and generating an erythema grading classification for erythema regions in psoriasis using the Skellam distribution statistics as feature vectors.Type: ApplicationFiled: December 5, 2014Publication date: June 9, 2016Inventor: Mithun Das Gupta
-
Publication number: 20160163041Abstract: Retinal vessel extraction on color images of eyes are disclosed. In one embodiment, the method comprises performing alpha matting on a multi-dimensional feature set derived from image data of a first image and performing retinal vessel extraction, including performing segmentation on the image by separating foreground and background image data from the first image using an output from the alpha matting.Type: ApplicationFiled: December 5, 2014Publication date: June 9, 2016Inventors: Powel Talwar, Mithun Das Gupta
-
Publication number: 20150327766Abstract: Embodiments segment received image data. The received image data may include pixels that have multiple channels of intensity values. The image data is decomposed into albedo and shading components. This may be accomplished using a minimization that enforces a relationship between albedo, shading, and intensity values. The minimization may also include an albedo regularizer to infer albedo in part based on chromaticity and albedo of surrounding pixels. Superpixels are generated based on contiguous regions of pixels having similar image data across channels. These superpixels are then merged based in part on the determined albedo and shading components as well as based on the image data. The channels of image data may include infrared image data used to modify visible channels of the image to create pseudo-image data, which may be used in place of image data for albedo-shading decomposition, superpixel generation, or superpixel merging.Type: ApplicationFiled: September 19, 2014Publication date: November 19, 2015Applicant: RICOH CO., LTD.Inventors: Mithun Das Gupta, Srinidhi Srinivasa
-
Patent number: 9147258Abstract: Methods and systems for segmentation in echocardiography are provided. One method includes obtaining echocardiographic images and defining a search space within the echocardiographic images using a pair of one-dimensional (1D) profiles. The method also includes using an energy based function constrained by non-local temporal priors within the defined search space to automatically segment a contour of a cardiac structure with the 1D profiles.Type: GrantFiled: February 26, 2013Date of Patent: September 29, 2015Assignee: General Electric CompanyInventors: Sheshadri Thiruvenkadam, Navneeth Subamanian, Mithun Das Gupta
-
Publication number: 20150254401Abstract: Embodiments digitize radiology films into DICOM format. Radiology films typically include an array of captured images laid out in a grid pattern. To comply with DICOM format, the scanned image of a radiology film is segmented into sub-images and text is extracted from the sub-images to generate DICOM metadata. The sub-images and extracted text metadata are then combined to generate a DICOM-compliant multi-image file.Type: ApplicationFiled: March 6, 2014Publication date: September 10, 2015Applicant: Ricoh Co., Ltd.Inventors: Pritesh Sankhe, Mithun Das Gupta
-
Patent number: 8891881Abstract: A method for identifying an optimal image frame is presented. The method includes receiving a selection of an anatomical region of interest in an object of interest. Furthermore, the method includes obtaining a plurality of image frames corresponding to the selected anatomical region of interest. The method also includes determining a real-time indicator corresponding to the plurality of acquired image frames, wherein the real-time indicator is representative of quality of an image frame. In addition, the method includes communicating the real-time indicator to aid in selecting an optimal image frame. Systems and non-transitory computer readable medium configured to perform the method for identifying an optimal image frame are also presented.Type: GrantFiled: January 25, 2012Date of Patent: November 18, 2014Assignee: General Electric CompanyInventors: Mithun Das Gupta, Kajoli Banerjee Krishnan, Pavan Kumar Veerabhadra Annangi, Xiaoming Liu, Sri Kaushik Pavani, Navneeth Subramanian, Jyotirmoy Banerjee