International Business Machines Corporation

Shanghai Tianma Micro-electronics Co., Ltd.

Tianma Micro-electronics Co., Ltd.

Shanghai Avic Opto Electronics Co., Ltd.

Yuan Yuan Ding

Peng Ji

Sheng Nan Zhu

Guo Qiang Hu

Fan Li

Hong Bing Zhang

Jian Ding

Jing Chang Huang

Dan Zhang

Yi Chen Zhong

Tian Tian Chai

Ke Zhang

Kerui Xi

Feng Qin

Yingteng Zhai

Tingting Cui

Yang Liu

Lei Niu

Jialing Li

Xue Ping Liu

Guoqiang Hu

Jine Liu

Xiaohe Li

Yongxin He

JingChang Huang

Xu Min

Chao Xin

Nuo Xu

Jun Ma

Yang Zeng

Zhong Fang Yuan

Tianyi Wu

Tong Liu

Yue Chen

Ying Liu

Jian Xu

Liang Xing

Jing Zhang

Hai Jun Xu

Jinfeng Li

Wenjun Dai

Yaoyang Liu

Donghua Chen

Hui Gao

Ting Cao

Si Tong Zhao

Yanfeng Shi

Huai Nan Zhou

Dong Chang

Chaomin Gao

Guancheng Xiao

Ziyi Jiang

Techniques for skin-condition visualization using machine learning. A color image depicting facial skin of a subject is retrieved. A monochromatic version of the color image is generated. Candidate instances of one or more skin conditions are segmented from the monochromatic version based on a segmentation threshold and using a machine learning model. A polarized version of the color image is generated, and based on the polarized version, the candidate instances are filtered. One or more simulation images are generated based on the filtered candidate instances.

Simulating progression of skin conditions based on machine learning

According to one embodiment, a method, computer system, and computer program product for building an acoustic model is provided. The present invention may include performing contrastive pre-training of the acoustic model; building a dataset classifier using prompt engineering; performing a prediction process; and performing zero-shot audio prediction using the pre-trained acoustic model.

Adaptable acoustic model built with limited labeling data

A display panel, a display device and a method for driving a display panel are provided. The display panel includes N types of display areas which includes an i-th type display area and a j-th type display area. The display panel includes M display parts which include a first display part and a second display part. The first display part includes at least one i-th type display area, and the second display part includes at least one i-th type display area. At least one j-th type display area is arranged between the i-th type display area included in the first display part and the i-th type display area included in the second display part. Light-emitting time periods of the i-th type display area and the j-th type display area at least partially do not overlap, to reduce the number of sub-pixels driven at the same time period.

Display panel and driving method for the same, and display device

According to one embodiment, a method, computer system, and computer program product for predicting the movement intentions of ships is provided. The present invention may include obtaining binocular camera visual data of a body of water; identifying physical objects in the body of water; generating feature point clouds of the physical objects in the body of water; performing instance segmentation on the generated feature point clouds; analyzing the generated feature point clouds; predicting the movement intentions of identified ships in the body of water based on the analyzed feature point clouds; and displaying the predicted movement intentions of the identified ships in the body of water on an augmented reality device.

AR and deep learning intersection ship route predictor

A plurality of different images of a same region of interest in an object are input into a set of neural networks, wherein each image of the region has been captured under a different value of a variable condition. A classification for each image is generated by the set of neural networks, wherein each classification includes a confidence score in a prediction of whether a feature is present in the region. The image classifications are ensembled to generate a final classification for the region. By applying a loss function, a loss is computed based on comparing the final classification to a ground truth of whether the feature is present in the region. The parameters of the set of neural networks are adjusted based on the computed loss.

Feature detection based on neural networks

In an aspect for generating a 2-dimensional (2D) building mapping, a computer-implemented method may include one or more processors configured for receiving scanning data in a first file format corresponding to 3-dimensional (3D) aerial Light Detection and Ranging (LiDaR) images. Further, the method may include transforming the first file format to a second file format and transmitting the scanning data to a building semantic segmentation component. Further, the computer-implemented method may include generating semantic segmentation data corresponding to building structures defined by point clusters; generating instance segmentation data corresponding to top surfaces of the building structures; generating building boundary data for each one of the point clusters based on one or more of the semantic segmentation data and the instance segmentation data; and processing the building boundary data to generate a 2D building mapping comprising refined boundary lines for each of the point clusters.

Generating 2D mapping using 3D data

A method for restoration of a kinetic event using video. The method includes obtaining video involving the kinetic event including a plurality of frames. An absolute location of a frame of the plurality of frames is determined. Movement of a feature point cloud corresponding to an object in motion in the plurality of frames is analyzed. A 3D object model with substantially similar dimensions to the object in motion is selected. The 3D object model is displayed emulating the analysed movement on the 3D map based on the absolute location.

Restoration of a kinetic event using video

Methods for creating a knowledge graph for a video are disclosed. Aspects include obtaining the video, processing the video to extract audio information and video information, and storing the extracted audio information and video information with a timestamp corresponding its occurrence in the video. Aspects also include creating a plurality of groups of the extracted audio information and video information based at least in part on the timestamps and extracting two or more keywords from each of the plurality of groups. Aspects further include identifying a relationship between the two or more keywords based on the extracted audio information and video information and creating a graph having a plurality of nodes and a plurality of links that connect a pair of nodes of the plurality of nodes. Each node corresponds to an extracted keyword and each link corresponds to the identified relationship between the pair of nodes.

Creating a knowledge graph for a video

An approach is provided in which the approach receives an image that includes multiple image points and constructs a plane in the image based on a first subset of the plurality of image points. The approach identifies a second subset of the image points that belong to the plane and are not part of the first subset of image points, and removes the first subset of image points and the second subset of image points form the image points. The approach annotates the remaining subset of image points in the image.

Three-dimensional segmentation annotation

A method for detecting and labeling a target object in a 2D image includes receiving a plurality of 2D images from a visual sensor, manually marking points of the target object on each of the 2D images, generating from the 2D images a 3D world coordinate system of the environment surrounding the target object, mapping each of the marked points on the 2D images to the 3D world coordinate system using a simultaneous localization and mapping (SLAM) engine, automatically generating a 3D bounding box covering all the marked points mapped to the 3D world coordinate system, mapping the 3D bounding box to each of the 2D images, generating a label for the target object on each of the 2D images using a machine learning object detection model, and training the machine learning object detection model based on the generated label for the target object.

AR-based labeling tool for 3D object detection model training

A system, method, and computer program product provides a way to separate connected or adhered adjacent characters of a digital image for license plate recognition. As a threshold processing, the method performs a recognition of character adhesion by obtaining character parameters using an image processor. The parameters include a horizontal max crossing and a ratio of width and height. A first rule-based module is used responsive to the character parameters to distinguish the adhered characters (character adhesions) that are easy to judge, leaving the uncertain part to a character adhesion classifier model for discrimination. Character adhesion data is obtained by data augmentation including the adding of a random distance between two single characters to create class like adhered characters. Then the character adhesion classifier model of single character and character adhesion data is trained. Any uncertain part can be distinguished by the trained character adhesion classifier model.

Character recognition of license plate under complex background

Provided are a display panel, a preparation method thereof and a display device. The display panel includes a substrate; a driving circuit layer, where the driving circuit layer is disposed on a side of the substrate, and includes a plurality of driving circuits; and an LED element layer, where the LED element layer is disposed on a side of the driving circuit layer facing away from the substrate, and includes a plurality of micro-LED elements, where each of the plurality of micro-LED elements is electrically connected to a respective one of the plurality of driving circuits through a via hole.

Display panel, preparation method thereof and display device

The present disclosure relates to training a machine learning model to classify images. An example method generally includes receiving a training data set including images in a first category and images in a second category. A convolutional neural network (CNN) is trained using the training data set, and a feature map is generated from layers of the CNN based on features of images in the training data set. A first area in the feature map including images in the first category and a second area in the feature map where images in the first category overlap with images in the second category are identified. The first category is split into a first subcategory corresponding to the first area and a second subcategory corresponding to the second area. The CNN is retrained based on the images in the first subcategory, images in the second subcategory, and images in the second category.

Classifying images in overlapping groups of images using convolutional neural networks

A method, a device and a computer program product for image processing are proposed. In the method, a first training image and region information are obtained. The region information indicates a region of a defect in the first training image. A second training image with the defect at least partially removed is generated using an image generator based on the first training image and the region information. The image generator is trained to recover the first training image by replacing pixels included in the region indicated by the region information. The image generator is updated based on the second training image. In this way, the image including the defect can be accurately and efficiently recovered.

Image recovery

Growing community of inventors

Shanghai, China

Yuan Yuan Ding