Provided is an information processing apparatus including: an obtaining unit configured to obtain a token string generated based on character strings included in a document image; a first determination unit configured to determine a document type represented by the document image and character strings corresponding to a first item included in the document image by using a result obtained by inputting the token string into a trained model; and a second determination unit configured to determine a character string corresponding to a second item by applying the document type and the character strings corresponding to the first item to a rule-based algorithm.

Computer vision systems and methods for end-to end training of neural networks are provided. The system generates a fixed point algorithm for dual-decomposition of a maximum-a-posteriori inference problem and trains the convolutional neural network and a conditional random field with the fixed point algorithm and a plurality of images of a dataset to learn to perform semantic image segmentation. The system can segment an attribute of an image of the dataset by the trained neural network and the conditional random field.

A character recognition device includes a recognizer that recognizes at least one character string from an image including a trailer captured by an imaging device, an attribute determinator that determines an attribute of the character string recognized by the recognition unit, and a trailer ID estimator that estimate whether the character string is a trailer ID based on the attribute of the character string determined by the attribute determinator.

An interaction checking method, device, and system determines health interactions, in real time, as images are collected from an environment and provides augmented informational data presentations of the same. The interactions are based on health information, historical information, and stored interaction data. In particular, the images are received from a camera viewing the environment and an identity is determined of the objects in the environment. Based on the identity determined, information about the objects is retrieved from a memory device, and an interaction warning is determined for a specific combination of the objects if the objects were to be used together. A display device augments visual data of the environment with the information about the objects in a presentation to a user that includes the interaction warning and information about the objects.

This disclosure is directed to, in part, mobile carts that are configured to determine their respective locations based on analysis of image data generated by cameras mounted to the respective carts. For instance, an example mobile cart may include at least one camera of a field-of-view directed substantially away from a cart and substantially towards an outward environment of the cart, such as toward an inventory location that houses one or more items. The mobile cart may generate image data representative of items housed at an inventory location adjacent to the cart and may use computer-vision techniques to analyze the image data and determine characteristics of these items. The mobile cart may then use this information to determine which section of multiple sections of a store in which the cart is currently located.

An information processing apparatus according to an embodiment of the present technology includes a classification unit and a generation unit. The classification unit classifies an object detected in a space on a basis of a predetermined criterion. The generation unit sets a priority for the object on a basis of a classification result by the classification unit, and generates position-related information regarding a position in the space on a basis of the set priority. Use of the position-related information makes it possible to improve the accuracy of autonomous movement control. This makes it possible to improve the accuracy of autonomous movement control.

Computer vision systems and methods for end-to end training of neural networks are provided. The system generates a fixed point algorithm for dual-decomposition of a maximum-a-posteriori inference problem and trains the convolutional neural network and a conditional random field with the fixed point algorithm and a plurality of images of a dataset to learn to perform semantic image segmentation. The system can segment an attribute of an image of the dataset by the trained neural network and the conditional random field.

An imaging part obtains an image of a read object, and a character string recognizing part recognizes a character string in the image. An output format setting part sets one or more output formats of a character string to be read from the image and to be output. A character extracting part obtains a character string for output, at a portion matching any of the one or more output formats among the recognized character string. At this time, a notifying part notifies a possibility of misreading in a case where the character string for output having characters less than a notification threshold number is obtained.

An imaging part obtains an image of a read object, and a character string recognizing part recognizes a character string in the image. An output format setting part sets an output format of a character string to be read from the image and to be output. A character extracting part obtains a candidate for a character string for output, at a portion matching the output format among the recognized character string, and if plural candidates are obtained, selects one of the plural candidates as the character string for output, based on a predetermined condition. A notifying part notifies a possibility of misreading in a case where plural candidates are obtained by the character extracting part.

An imaging part obtains an image of a read object, a shape recognizing part recognizes shapes in the image, and a character string recognizing part recognizes a character string among the recognized shapes. A notifying part notifies a possibility of misreading in a case where a shape, among the recognized shapes, not constituting any character exists in or near the recognized character string. When a character extracting part obtains a character string for output, that is all or a part of the recognized character string, the notification of the possibility of misreading may be performed also in a case where a shape not constituting any character exists in or near the obtained character string for output.