Koninklijke Philips Corporation N.v.

Microsoft Technology Licensing, LLC

Siemens Aktiengesellschaft

Vipin Gupta

Pallavi Vajinepalli

Payal Keswarpu

Celine Firtion

Shankar Mosur Venkatesan

Hari Sundar

Amit Kale

Caifeng Shan

Marinus Bastiaan Van Leeuwen

Jelte Peter Vink

Ananth Sankar

Manfred Mueller

Subhendu Seth

Lu Wang

Saurabh Kataria

Sarif Kumar Naik

Ying Xuan

Sushanth Govinahallisathyanarayana

Bharat Kumar Jain

Subhash Chandra Bose Gali

Shubham Agarwal

Mandar Shirikant Kulkarni

Bhargavkumar Kanubhai Patel

Sirjan Kafle

Di Wen

Sakshi Verma

Xue Xia

Nikita Gupta

Xun Luan

Aman Gupta

Described herein are systems and methods for generating an embedding—a learned representation—for an image. The embedding for the image is derived to capture visual aspects, as well as textual aspects, of the image. An encoder-decoder is trained to generate the visual representation of the image. An optical character recognition (OCR) algorithm is used to identify text/words in the image. From these words, an embedding is derived by performing an average pooling operation on pre-trained embeddings that map to the identified words. Finally, the embedding representing the visual aspects of the image is combined with the embedding representing the textual aspects of the image to generate a final embedding for the image.

Utilizing visual and textual aspects of images with recommendation systems

Technologies for evaluating, scoring, and determining whether to present a caption of an image are provided. The disclosed techniques include receiving an image with associated metadata. Contextual data is identified from the image and the metadata. A generated caption for the image is received from an image caption generator. A first vector representation is generated based on the contextual image data and a second vector representation is generated based on the generated caption. A machine learned model generates a score for the generated caption using the first vector representation and the second vector representation. The score represents a confidence value defining how accurately the caption describes the image. Based on the score, the caption may be presented along with the image on a client device.

Automatically evaluating caption quality of rich media using context learning

Methods, systems, and computer programs are presented for detecting near-duplicate profile images of the users in a social network. One method includes operations for identifying an image in a profile of a user of the social network, determining a query feature vector for the image, the query feature vector comprising a set of features, and determining a dominant feature from the features, the dominant feature having a highest value from the values of the features. Further, the method includes operations for determining a bucket in a database of feature vectors based on the dominant feature, determining if the query feature vector is a near duplicate of any feature vector in the determined bucket, and determining if the profile of the user is a duplicate profile or a fake profile based on whether the query feature vector is a near duplicate of any feature vector in the determined bucket.

Detection of near-duplicate images in profiles for detection of fake-profile accounts

The invention relates to a photonic probe apparatus and a method for probing tissue to detect and mark biological tissue with cancerous or precancerous states. The apparatus involves a probe for illuminating tissue and collecting light from an illuminated tissue region through the probe, a unit for analyzing collected light to determine whether a threshold measure of probability of a cancerous or precancerous lesion in the probed tissue region in contact with the probe is exceeded, and an integrated tissue marking facility which can be activated to mark the probed tissue region through the probe when the threshold measure is exceeded. The photonic probe apparatus and the method are especially suitable for probing regions in squamous and columnar epithelia to detect and mark regions with cervical cancer or cervical intraepithelial neoplasia (CIN).

Photonic probe apparatus with integrated tissue marking facility

The invention provides a system () for acquiring cervical images which comprises an image acquisition subsystem () for acquiring cervical images () of a cervical region of a patient during a colposcopy procedure, and a display subsystem () for displaying the cervical images on a display () by providing image data () of the cervical images to the display. According to the invention, the image acquisition subsystem () is arranged for, when operating in an interval mode, acquiring the cervical images () at predetermined time intervals to obtain a time-series of cervical images showing changes in the cervical region over time, and reporting a progress of said acquiring to the display subsystem. Moreover, the display subsystem () is arranged for establishing a progress indicator (-) on the display by generating indicator data () and providing the indicator data to the display (), the progress indicator providing visual feedback on the progress of said acquiring to a user. Accordingly, the user is provided with visual feedback on a status of the acquiring of the cervical images. In particular, the progress indicator avoids that the user experiences a loss of control or confusion.

Acquiring cervical images

A system () is provided for imaging a patient's interior. The system comprises an imaging sensor () for acquiring a series of images of a region of interest () in the patient's interior, and a plurality of light sources () for illuminating the region of interest in the patient's interior from different light source directions (-). A light controller () is provided for controlling individual ones (-) of the plurality of light sources () to dynamically vary the light source directions (-) during said acquiring. Moreover, a processor () is provided for obtaining lighting data () indicative of the dynamically varying light source directions. The processor () is further arranged for using the lighting data to apply a photometric stereo technique to the image data so as to establish a three-dimensional [3D] surface profile of the region of interest. The processor () is further provided to detect insufficiently illuminated areas of the region of interest (-) in the images acquired by the imaging sensor (), using the 3D surface profile of the region of interest. Accordingly, the system is enabled to establish a 3D surface profile of the region of interest from a series of images in a convenient and cost-effective manner.

Imaging a patient's interior

The invention provides for a medical instrument for examining the cervix comprising an optical examination system, a processor for controlling the medical instrument, and a memory containing machine executable instructions. Execution of the instructions causes the processor to: acquire a cervical image using the optical examination system; calculate a set of interest point locations using a digital filter; calculate a filtered set of interest point locations using the set of interest point locations and a morphological filter; calculate a reduced set of interest points locations using the filtered set of interest point locations and a neighborhood based filter; calculate a classified set of interest point locations reduced set of interest points and a trained classification module; calculate a set of punctation locations using the classified set of interest point locations and a second neighborhood based filter; and calculate punctation mark regions using the punctation point locations.

Medical instrument for examining the cervix

A medical imaging system to segment an original blood vessel of a body part represented by an original medical image is provided. The system includes an image analyzer for receiving and the original medical image to analyze the original medical image to provide a Hessian Eigen analysis comprising a first data and a second data mapped to each pixel of the medical image, and an image identifier for receiving the Hessian Eigen analysis and for identifying seed points from the pixels by processing the first data and the second data along with a vesselness property, wherein the seed points are used for segmenting the original blood vessel to provide a corrected medical image representing a corrected blood vessel.

Medical imaging system for segementing blood vessel

A medical imaging system for processing an original medical image is provided. The system includes a seed generator for receiving the original medical image representing an original blood vessel and to generate coarse seeds on a basis of vesselness feature in respect to every pixel of the original blood vessel represented by the medical image, and a seed processor for receiving the coarse seeds and for processing the coarse seeds to select a set of refined seeds on a basis of the regression profile for each of the coarse seed by using random forest classification, such that the set of refined seeds are selected from the coarse seeds such that the set of refined seeds adapted to lie on a corrected blood vessel.

Medical imaging system

Growing community of inventors

Bengaluru, India

Vipin Gupta