摘要:
A method and system for inspecting railway components. The method includes receiving a stream of images containing railway components, detecting a railway component in each image, generating a plurality of feature vectors for each railway component image, measuring the dissimilarity between the railway component and a set of railway components detected in preceding images, in a sliding window, based on the feature vectors.
摘要:
View-specific object detectors are learned as a function of scene geometry and object motion patterns. Motion directions are determined for object images extracted from a training dataset and collected from different camera scene viewpoints. The object images are categorized into clusters as a function of similarities of their determined motion directions, the object images in each cluster are acquired from the same camera scene viewpoint. Zenith angles are estimated for object image poses in the clusters relative to a position of a horizon in the cluster camera scene viewpoint, and azimuth angles of the poses as a function of a relation of the determined motion directions of the clustered images to the cluster camera scene viewpoint. Detectors are thus built for recognizing objects in input video, one for each of the clusters, and associated with the estimated zenith angles and azimuth angles of the poses of the respective clusters.
摘要:
A method and system is provided to provide enrollment of biometric data from individuals without the need for the enrollees to travel to a central location. The system and method provides for optionally detecting multiple types of biometric data, for example, fingerprints, facial scan, visual scan, iris scan, voice scan, or the like, to be captured at the point of enrollee use. During the biometric capture process, the biometric enrollment device monitors and establishes the identity of the enrollee and measures the quality of the biometric input as it occurs. If unacceptable quality is detected, then repeated biometric scans may be necessary, but is done at time of enrollment avoiding any need to send inaccurate information back to a service institution. When the enrollment process completes, the biometric data is encrypted, time stamped and either mailed back to the service institution or is transmitted back. The biometric device may remain with the user for subsequent use as an identification sensor which may authenticate a user with stored biometric information to authenticate a user's identity for a transaction or a request to access a service or equipment. The biometric enrollment device may also be embodied with another piece of equipment to authenticate use of that equipment.
摘要:
Training data object images are clustered as a function of motion direction attributes and resized from respective original into same aspect ratios. Motionlet detectors are learned for each of the sets from features extracted from the resized object blobs. A deformable sliding window is applied to detect an object blob in input by varying window size, shape or aspect ratio to conform to a shape of the detected input video object blob. A motion direction of an underlying image patch of the detected input video object blob is extracted and motionlet detectors selected and applied that have similar motion directions. An object is thus detected within the detected blob and semantic attributes of an underlying image patch extracted if a motionlet detectors fires, the extracted semantic attributes available for use for searching for the detected object.
摘要:
A method, data processing system, and computer program product for managing a headlight on a vehicle are presented. An image of an area in front of the vehicle is received. A first set of features is identified in the received image. The first set of features in the received image is compared with a number of sets of features from a plurality of previous images. Each image in the plurality of previous images is associated with a headlight setting. A second set of features from a previous image in the plurality of previous images matching the first set of features in the received image is identified. A determination is made whether to change a setting for the headlight on the vehicle based on the headlight setting associated with the previous image.
摘要:
The present invention provides an approach for determining a state of a shopping receptacle. Specifically, under the present approach, an image capture device (e.g., camera) is positioned at the checkout station and/or on the shopping receptacle. The device will capture at least one image of the shopping receptacle during the checkout process and/or when the customer or checkout clerk indicates that the transaction is ready for payment (e.g., all items have been scanned or rung through). Using such image(s), a model for an empty receptacle and a knowledgebase of non-shopping items that may remain in the shopping receptacle, it is determined whether the receptacle is empty. If not, an alarm or other notification will be activated and the events can be logged.
摘要:
Under the present invention, a single, overall alarm for an entire set of shopping items will be used for any and all discrepancies. The metric used for creating an alarm for the overall set of shopping items can be based on any one of the following candidate policies: if at least one item generated an alarm; if some fixed number of items generated an alarm; if some threshold discrepancy metric got exceed; if basket size is larger than certain threshold cash value and the alarm exceeded certain threshold alarm rate; a randomly generated alarm (e.g., random audit); the customer's identity and track record (e.g., loyalty card); and/or any combination of the above. Regardless, if an overall alarm is generated one or more of the following actions can be taken: no action send the customer to customer service; appropriately record customer track record (e.g., loyalty card) when customer identity is available; audit the customer at the “shop exit; and/or any combination of thereof.
摘要:
Under the present invention, item verification is automated and expedited. Specifically, items to be purchased can be scanned by the shopper using a barcode reader (e.g., a scanner), attached to or positioned near the checkout station. As items are scanned, they are identified based on their barcode, and added to an item list. Item verification can then be performed at checkout using imaging technology. Specifically, as items are scanned, an item verification unit will capture an appearance thereof (via a camera). Item verification software within the item verification unit will access a database that associates items with their images/appearances. The appearance will be compared for consistency to the identity as determined based on the scan. In general, the item verification unit is a separate unit from the cash register, but adapted to work in conjunction therewith (e.g., as a pluggable system, via wireless communication, etc.).
摘要:
Foreground objects of interest are distinguished from a background model by dividing a region of interest of a video data image into a grid array of individual cells that are each smaller than that a foreground object of interest. More particularly, image data of the foreground object of interest spans a contiguous plurality of the cells. Each of the cells are labeled as foreground if accumulated edge energy within the cell meets an edge energy threshold, if color intensities for different colors within each cell differ by a color intensity differential threshold, or as a function of combinations of said determinations in view of one or more combination rules.
摘要:
Images are retrieved and ranked according to relevance to attributes of a multi-attribute query through training image attribute detectors for different attributes annotated in a training dataset. Pair-wise correlations are learned between pairs of the annotated attributes from the training dataset of images. Image datasets may then be searched via the trained attribute detectors for images comprising attributes in a multi-attribute query, wherein images are retrieved from the searching that each comprise one or more of the query attributes and also in response to information from the trained attribute detectors corresponding to attributes that are not a part of the query but are relevant to the query attributes as a function of the learned plurality of pair-wise correlations. The retrieved images are ranked as a function of respective total numbers of attributes within the query subset attributes.