Approaches introduce a pre-processing and post-processing framework to a neural network-based approach to identify items represented in an image. For example, a classifier that is trained on several categories can be provided. An image that includes a representation of an item of interest is obtained. Rotated versions of the image are generated and each of a subset of the rotated images is analyzed to determine a probability that a respective image includes an instance of a particular category. The probabilities can be used to determine a probability distribution of output category data, and the data can be analyzed to select an image of the rotated versions of the image. Thereafter, a categorization tree can then be utilized, whereby for the item of interest represented the image, the category of the item can be determined. The determined category can be provided to an item retrieval algorithm to determine primary content for the item of interest. This information also can be used to determine recommendations, advertising, or other supplemental content, within a specific category, to be displayed with the primary content.

An obstacle warning method includes the steps of: acquiring scenario images at a current sampling time and a previous sampling time, and a map about first relative distances between respective viewing points in a viewing field and a vehicle; acquiring a profile and marking information of an obstacle and a map about a second relative distance between the obstacle and the vehicle in accordance with the map about the first relative distances; calculating a map about a third relative distance between the obstacle and the vehicle at a previous sampling time in accordance with the map about the first relative distances at the previous sampling time, the profile and the marking information of the obstacle at the current sampling time and a motion vector of the obstacle from the current sampling time to the previous sampling time.

An image search device includes: a search unit that retrieves an image similar to a search target image from a registration unit where an image and link information are registered in an associated manner; and a transmission unit that transmits, to a terminal device, the link information associated with the retrieved image. An information processing apparatus controls registration of an image and link information in the registration unit. The information processing apparatus extracts a predetermined area from a registration target image to be registered in the registration unit, transmits an image of the predetermined area to the search unit, and determines whether a similar image including an image similar to the predetermined area is registered in the registration unit based on a search result of the search unit. In the case where determination is made that a similar image is registered in the registration unit, such a fact is notified.

A rectangular region group storage unit stores a group of rectangular regions indicating portions to be recognized for a crowd state on an image. A crowd state recognition dictionary storage unit stores a dictionary of a discriminator acquired by machine learning by use of a plurality of pairs of crowd state image as an image which expresses a crowd state at a predetermined size and includes a person whose reference site is expressed as large as the size of the reference site of a person defined for the predetermined size, and training label for the crowd state image. A crowd state recognition unit extracts regions indicated in the group of rectangular regions stored in the rectangular region group storage unit from a given image, and recognizes states of the crowds shot in the extracted images based on the dictionary.

Embodiments of the present invention disclose a method and an apparatus of establishing an image search relevance prediction model, and an image search method and apparatus. The method of establishing an image search relevance prediction model comprises: training a pre-constructed original deep neural network by using a training sample, wherein the training sample comprises: a query and image data, and the original deep neural network comprises: a representation vector generation network and a relevance calculation network; and using the trained original deep neural network as the image search relevance prediction model. The technical solution of the present invention optimizes the existing image search technology, and has stronger capabilities than the prior art as well as various integrations and variations in terms of semantic matching between a query and an image text, semantic matching between a query and image content, click generalization and the like.

A method and system for controlling a home security system. A processor may be trained to recognize an image standard for a scene, wherein the training comprises creating a profile of the image standard. Operational imaging of the scene may be performed to create an operational image. A profile of the operational image may be created. Profiles of the image standard and the operational image may be compared. A state of a security system may be changed as a result of a comparison of the profiles of the image standard and the operational image.

A method for generating object and part detectors includes accessing a collection of training images. The collection of training images includes images annotated with an object label and images annotated with a respective part label for each of a plurality of parts of the object. Joint appearance-geometric embeddings for regions of a set of the training images are generated. At least one detector for the object and its parts is learnt using annotations of the training images and respective joint appearance-geometric embeddings, e.g., using multi-instance learning for generating parameters of scoring functions which are used to identify high scoring regions for learning the object and its parts. The detectors may be output or used to label regions of a new image with object and part labels.

Disclosed embodiments provide for deep convolutional computing image analysis. The convolutional computing is accomplished using a multilayered analysis engine. The multilayered analysis engine includes a deep learning network using a convolutional neural network (CNN). The multilayered analysis engine is used to analyze multiple images in a supervised or unsupervised learning process. The multilayered engine is provided multiple images, and the multilayered analysis engine is trained with those images. A subject image is then evaluated by the multilayered analysis engine by analyzing pixels within the subject image to identify a facial portion and identifying a facial expression based on the facial portion. Mental states are inferred using the deep convolutional computer multilayered analysis engine based on the facial expression.

An information processing apparatus includes a receiving unit, a detecting unit, and an associating unit. The receiving unit receives a captured moving image of a target person. The detecting unit detects a cue given by the target person. The associating unit associates the cue, given by the target person and detected by the detecting unit, with the moving image so that the cue will be used to designate the moving image used for evaluating the target person.

A transformed image is received. The transformed image includes an other-than-visible light image that has been captured using a transformation device. A region of the transformed image is isolated, the region being less than an entirety of the transformed image. By applying to the region a convolutional Neural Network (CNN) which executes using a processor and a memory, and by processing only the region of the transformed image, an object of interest is detected in the region. Upon detecting, an indication is produced to indicate the presence of the object of interest in the region.