摘要:
One or more techniques and/or systems are disclosed for detecting a quick response (QR) code. An area of an image comprising a QR code is localized by combining pixel dynamic scale (DS), black-cell ratio (BR), and edge intensity sum (EIS) criteria determination to identify the QR code. A pattern for the QR code is detected, comprising determining if a position detection pattern (PDP) is located in respective grid areas of a first grid that comprises the QR code, and identifying an alignment pattern (AP), if present. To identify the AP, an AP region is estimated using the PDPs, and a center area of the AP is found by examining respective areas of a second grid comprising the estimated AP region.
摘要:
Visual concepts contained within a video clip are classified based upon a set of target concepts. The clip is segmented into shots and a multi-layer multi-instance (MLMI) structured metadata representation of each shot is constructed. A set of pre-generated trained models of the target concepts is validated using a set of training shots. An MLMI kernel is recursively generated which models the MLMI structured metadata representation of each shot by comparing prescribed pairs of shots. The MLMI kernel is subsequently utilized to generate a learned objective decision function which learns a classifier for determining if a particular shot (that is not in the set of training shots) contains instances of the target concepts. A regularization framework can also be utilized in conjunction with the MLMI kernel to generate modified learned objective decision functions. The regularization framework introduces explicit constraints which serve to maximize the precision of the classifier.
摘要:
Method, system, and programs for providing identifiers to objects. Input data representing a plurality of objects is received and categorized into a plurality of entity categories. A first graph of entities is generated using the plurality of entity categories. The first graph of entities are matched with a second graph of entities. A comparison of object pairs is then made, in which each object pair includes a first object from the first graph of entities and a corresponding second object from the second graph of entities. Identifiers are assigned to each object based on comparing the object pairs.
摘要:
Method, system, and programs for providing identifiers to objects. Input data representing a plurality of objects is received and categorized into a plurality of entity categories. A first graph of entities is generated using the plurality of entity categories. The first graph of entities are matched with a second graph of entities. A comparison of object pairs is then made, in which each object pair includes a first object from the first graph of entities and a corresponding second object from the second graph of entities. Identifiers are assigned to each object based on comparing the object pairs.
摘要:
Visual concepts contained within a video clip are classified based upon a set of target concepts. The clip is segmented into shots and a multi-layer multi-instance (MLMI) structured metadata representation of each shot is constructed. A set of pre-generated trained models of the target concepts is validated using a set of training shots. An MLMI kernel is recursively generated which models the MLMI structured metadata representation of each shot by comparing prescribed pairs of shots. The MLMI kernel is subsequently utilized to generate a learned objective decision function which learns a classifier for determining if a particular shot (that is not in the set of training shots) contains instances of the target concepts. A regularization framework can also be utilized in conjunction with the MLMI kernel to generate modified learned objective decision functions. The regularization framework introduces explicit constraints which serve to maximize the precision of the classifier.