A method and device for automatically classifying document hardcopy images by using document hardcopy image descriptors are provided. The method and device include providing a document hardcopy image, the document hardcopy image having image features, extracting image descriptors by a first set of image descriptor extractors, each image descriptor of the image descriptors being descriptive of the image features of the document hardcopy image, estimating class probabilities of the document hardcopy image by multiple trained classifiers based on the image descriptors, determining a most probable class of the document hardcopy image by a trained meta-classifier based on the class probabilities estimated by the multiple trained classifiers, inputting the document hardcopy image and the most probable class of the document hardcopy image to an assigner, and assigning, by the assigner, the most probable class determined by the trained meta-classifier to the document hardcopy image to obtain a classified document hardcopy image.

Provided is a behavior recognition apparatus, including a detection unit configured to detect, based on a vehicle interior image obtained by photographing a vehicle interior, positions of a plurality of body parts of a person inside a vehicle in the vehicle interior image; a feature extraction unit configured to extract a rank-order feature which is a feature based on a rank-order of a magnitude of a distance between parts obtained by the detection unit; and a discrimination unit configured to discriminate a behavior of an occupant in the vehicle using a discriminator learned in advance and the rank-order feature extracted by the feature extraction unit. Also provided is a learning apparatus to learn the discrimination unit.

Embodiments include a system, method, and computer program product for distributed intelligent pattern recognition. Embodiments include a cooperative multi-agent detection system that enables an array of disjunctive devices (e.g., cameras, sensors) to selectively cooperate to identify objects of interest over time and space, and to contribute an object of interest to a shared deep learning pattern recognition system based on a bidirectional feedback mechanism. Embodiments provide updated information and/or algorithms to one or more agencies for local system learning and pattern updating recognition models. Each of the multiple agencies may in turn, update devices (e.g., cameras, sensors) coupled to the local machine learning and pattern recognition models.

A system and method are provided for performing saliency detection on an image or video. The method includes training and creating deep features using deep neural networks, such that an input image is transformed into a plurality of regions, which minimizes intra-class variance, and maximizes inter-class variance, according to one or more active contour energy constraints. The method also includes providing and output associated with the deep features.

A system estimates quality of a digital image by accessing a corpus of digital images of one or more subjects, such as a facet of a property. The system will receive, for at least a subset of the corpus, an indicator that one or more patches of each image in the subset is out of focus. The system will train a classifier by obtaining a feature representation of each pixel in each image, along with a focus value that represents an extent to which each pixel in the image is in focus or out of focus. The system will use the classifier to analyze pixels of a new digital image and assess whether each analyzed pixel in the new digital image is in focus or out of focus. The system may use the image to assess whether an incident occurred, such as storm-related damage to the property.

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating numeric embeddings of images. One of the methods includes obtaining training images; generating a plurality of triplets of training images; and training a neural network on each of the triplets to determine trained values of a plurality of parameters of the neural network, wherein training the neural network comprises, for each of the triplets: processing the anchor image in the triplet using the neural network to generate a numeric embedding of the anchor image; processing the positive image in the triplet using the neural network to generate a numeric embedding of the positive image; processing the negative image in the triplet using the neural network to generate a numeric embedding of the negative image; computing a triplet loss; and adjusting the current values of the parameters of the neural network using the triplet loss.

The present disclosure provides an image capture, curation, and editing system that includes a resource-efficient mobile image capture device that continuously captures images. In particular, the present disclosure provides low power frameworks for processing, compressing, and transmitting images at a mobile image capture device. One example low power framework includes a scene analyzer that analyzes a scene depicted by a first image and determines whether to store the first image in a non-volatile memory or to discard the first image from a temporary image buffer without storing the first image in the non-volatile memory.

A method for providing gesture recognition services to a user application, comprising: storing sets of training data in a database at a server, the training data received from a sensor associated with the user application, the training data being indicative of characteristics of a gesture, the user application running on a client device; training a gesture recognition algorithm with the sets of training data to generate a trained gesture recognition algorithm, the output of the trained gesture recognition algorithm being an indication of the gesture; storing the trained gesture recognition algorithm in a client library at the server; receiving raw data from the sensor via the user application and storing the raw data in the client library; applying the trained gesture recognition algorithm to the raw data; and, when the trained gesture recognition algorithm recognizes the gesture, sending the indication of the gesture from the client library to the user application.

In a method and system for identifying objects in an image, an image and training data are received. The training data identifies a pixel associated with an object of a particular type in the image. A plurality of filtered versions of the image are developed. The training data and the plurality of filtered versions of the image are processed to develop a trained model for classifying pixels associated with objects of the particular type. The trained model is applied to the image to identify pixels associated a plurality of objects of the particular type in the image. Additional image processing steps are developed to further refine the identified pixels for better fitting of the contour of the objects with their edges.

Embodiments are provided to search for a dictionary image corresponding to a target image. The method includes detecting keypoints in a set of dictionary images. The set of dictionary images includes at least one dictionary image having a plurality of pixels. At least one random pair of pixels is selected among the detected keypoints of the dictionary image on the basis of candidate coordinates for pixels distributed around the detected keypoints of the dictionary image. A feature vector of each keypoint of the dictionary image is calculated, including calculating a difference in brightness between the selected pairs of pixels of the dictionary image. The calculated difference in brightness is an element of the feature vector. Keypoints of a target image are detected.