Carnegie Mellon University

Gm Global Technolgoy Operations, Inc.

Microsoft Technology Licensing, LLC

Industrial Technology Research Institute

Future Wei Technologies, Inc.

Tsuhan Chen

Wende Zhang

Qi Wu

Hong-Jiang Zhang

Fu Jie Huang

Hong Heather Yu

Dongqing Zhang

Deepak S Turaga

Yuh-Feng Hsu

Trista Pei-Chun Chen

Kuan-chuan Peng

A method for detecting a clear path of travel for a vehicle using a current image generated by a camera includes defining an exemplary vanishing point for each of a plurality of sample images. Features within each of the plurality of sample images are identified, the current image generated by the camera is monitored, and features within the current image are identified. The current image is matched to at least one of the sample images based upon the identified features within the current image and the identified features within the plurality of sample images. A vanishing point for the current image is determined based upon the matching and the exemplary vanishing point for each of the matched sample images. The vanishing point for the current image is utilized to define a clear path of travel, and the clear path of travel is utilized to navigate the vehicle.

Clear path detection using an example-based approach

Apparatus and method for modification of emotion conveyed by an image. In one aspect, a metric corresponding to the emotion conveyed by an image is generated, and a command for modifying the emotion of the image causes the emotion conveyed by the image to be changed according to the specified change to emotion. The command to modify the image emotion can be made via an emotion keyword, and can include an indicated magnitude of emotion change. Emotion conveyed by an image can be specified via an emotion vector, where elements of the vector correspond with emotions arranged along orthogonal dimensions in an emotion space. In one aspect, an emotion predictor is trained to predict the emotion conveyed by an image via determining characteristic values for a set of image feature categories, based on a set of images associated with emotion keywords of the emotion space.

Emotion modification for image and video content

A method for detecting a clear path of travel for a vehicle utilizing analysis of an image generated by a camera device located upon the vehicle includes monitoring the image, identifying through patch-based clear path detection analysis of the image a first patch within the image that indicates a not clear path, analyzing the first patch through patch smoothing, invalidating the first patch based upon the analyzing the first patch through patch smoothing, utilizing the invalidated first patch to define a clear path of travel for the vehicle, and utilizing the clear path of travel to navigate the vehicle.

Clear path detection with patch smoothing approach

A method for detecting a clear path of travel for a vehicle using a current image generated by a camera includes defining an exemplary clear path for each of a plurality of sample images, identifying features within each of the plurality of sample images, monitoring the current image generated by the camera, identifying features within the current image, matching the current image to at least one of the sample images based upon the identified features within the current image and the identified features within the plurality of sample images, determining a clear path of travel based upon the matching and the exemplary clear path for each of the matched sample images, and utilizing the clear path of travel to navigate the vehicle.

A method for detecting a clear path of travel for a vehicle utilizing an image generated by a camera device located upon the vehicle includes monitoring the image, and generating a plurality of patches onto the image, wherein the patches are collectively arranged to substantially include the clear path of travel. Feature extraction is utilized to analyze the patches and includes convolving each of the patches with a feature detection filter generating a feature-based filter response, extracting features based upon the feature-based filter response, and determining each of the patches to represent the clear path of travel or to not represent the clear path of travel based upon the extracted features. The clear path of travel is determined based upon the plurality of patches, and is utilized to navigate the vehicle.

Clear path detection using patch approach

A method for estimating a vanishing point in a roadway using a current image generated by a camera on a vehicle includes defining an exemplary vanishing point for each of a plurality of sample images, identifying features within each of the plurality of sample images, monitoring the current image generated by the camera, identifying features within the current image, matching the current image to at least one of the sample images based upon the identified features within the current image and the identified features within the plurality of sample images, determining a vanishing point based upon the matching and the exemplary vanishing points for each of the matched sample images, and utilizing the vanishing point to navigate the vehicle.

Clear path detection using a vanishing point

A method of error concealment that includes providing a first set of image data at a first time instant, providing a second set of image data at a second time instant, representing the first set of image data by a model of updating mixture of principal components (UMPC) to form a UMPC model having a plurality of components, receiving the second set of image data, wherein the second set of image data contains error-free data, and updating the UMPC model using the error-free data at the second time instant.

Method for video error concealment by updating statistics

A method for making a mode decision in video coding. The method includes collecting a first portion of video data and labeling the first portion of video data with an optimal mode. The method also includes identifying a feature of the first portion of video data corresponding to the optimal mode and making a mode decision for a second portion of the video data based on a value of the feature in the second portion of the video.

System and methods for video compression mode decisions

A face recognition system and process for identifying a person depicted in an input image and their face pose. This system and process entails locating and extracting face regions belonging to known people from a set of model images, and determining the face pose for each of the face regions extracted. All the extracted face regions are preprocessed by normalizing, cropping, categorizing and finally abstracting them. More specifically, the images are normalized and cropped to show only a persons face, categorized according to the face pose of the depicted person's face by assigning them to one of a series of face pose ranges, and abstracted preferably via an eigenface approach. The preprocessed face images are preferably used to train a neural network ensemble having a first stage made up of a bank of face recognition neural networks each of which is dedicated to a particular pose range, and a second stage constituting a single fusing neural network that is used to combine the outputs from each of the first stage neural networks. Once trained, the input of a face region which has been extracted from an input image and preprocessed (i.e., normalized, cropped and abstracted) will cause just one of the output units of the fusing portion of the neural network ensemble to become active. The active output unit indicates either the identify of the person whose face was extracted from the input image and the associated face pose, or that the identity of the person is unknown to the system.

Pose-invariant face recognition system and process

A face recognition system and process for identifying a person depicted in an input image and their face pose. This system and process entails locating and extracting face regions belonging to known people from a set of model images, and determining the face pose for each of the face regions extracted. All the extracted face regions are preprocessed by normalizing, cropping, categorizing and finally abstracting them. More specifically, the images are normalized and cropped to show only a person's face, categorized according to the face pose of the depicted person's face by assigning them to one of a series of face pose ranges, and abstracted preferably via an eigenface approach. The preprocessed face images are preferably used to train a neural network ensemble having a first stage made up of a bank of face recognition neural networks each of which is dedicated to a particular pose range, and a second stage constituting a single fusing neural network that is used to combine the outputs from each of the first stage neural networks. Once trained, the input of a face region which has been extracted from an input image and preprocessed (i.e., normalized, cropped and abstracted) will cause just one of the output units of the fusing portion of the neural network ensemble to become active. The active output unit indicates either the identify of the person whose face was extracted from the input image and the associated face pose, or that the identity of the person is unknown to the system.

Growing community of inventors

Ithaca, NY, United States of America

Tsuhan Chen