Abstract: A mobile device capable of offline and online synchronous image identifying is disclosed. The mobile device includes an image capturing unit, a processor, and a display monitor. The image capturing unit retrieves frames of a recognition target. The processor determines whether the mobile device is operatively connected to Internet. Also, the processor sends the frames to a recognition server via the Internet, and performs offline image recognition to the frames and displays the result of the offline image recognition on the display monitor when the mobile device is connected to the Internet, and receives a result of an online image recognition performed synchronously to the frames by the recognition server and displays the result of the online image recognition on the display monitor when the result of the online image recognition has a recognition value higher than a recognition value of the result of the offline image recognition.

Abstract: A system and a method for multi-modal fusion based fault tolerant video content recognition is disclosed. The method conducts multi-modal recognition on an input video to extract multiple components and their respective appearance time in the video. Next, the multiple components are categorized and recognized respectively via different algorithms. Next, when the recognition confidence of any component is insufficient, a cross-validation with other components is performed to increase the recognition confidence and improve the fault tolerance of the components. Furthermore, when the recognition confidence of an individual component is insufficient, the recognition continues and tracks the component, spatially and temporally when it applies, until frames of high recognition confidence in the continuous time period is reached.

Abstract: An image identifying method for offline and online synchronous operation is disclosed. When the mobile device focuses on an recognition target, images frames of an recognition target are retrieved and sent to an recognition server. The mobile device executes an offline recognition operation, and synchronously an recognition server executes an online recognition operation. A plurality of the matching data is saved in the mobile device, and a plurality of recognition data is saved in an recognition server, wherein the recognition data larger than the matching data. When the mobile device firstly generates the recognition result, the recognition result is displayed for user searching reference. If the mobile device receives an recognition result from the recognition server before the mobile device completing the recognition, the recognition result replied form the recognition server is displayed on the display monitor.

Abstract: A cumulative image recognition method performs a matching analysis on frame and matching data in a mobile device when the mobile device focuses on a recognition target and captures frame of the image of the recognition target. The method obtains the feature value of each matching data according to the matching result, and determines if one of the matching data has an optimal result matching the recognition target according to several entries of feature values. If no optimal result, then the method respectively sums up the feature value of each matching data into a summing feature value and determines if the plurality of the matching data has a candidate result similar to the recognition target according to several entries of the summing feature values. If no candidate result, the mobile device captures the next frame on the recognition target, and performs next matching analysis on the next frame.

Abstract: A cumulative image recognition method performs a matching analysis on frame and matching data in a mobile device when the mobile device focuses on a recognition target and captures frame of the image of the recognition target. The method obtains the feature value of each matching data according to the matching result, and determines if one of the matching data has an optimal result matching the recognition target according to several entries of feature values. If no optimal result, then the method respectively sums up the feature value of each matching data into a summing feature value and determines if the plurality of the matching data has a candidate result similar to the recognition target according to several entries of the summing feature values. If no candidate result, the mobile device captures the next frame on the recognition target, and performs next matching analysis on the next frame.

Abstract: An image identifying method for offline and online synchronous operation is disclosed. When the mobile device focuses on an recognition target, images frames of an recognition target are retrieved and sent to an recognition server. The mobile device executes an offline recognition operation, and synchronously an recognition server executes an online recognition operation. A plurality of the matching data is saved in the mobile device, and a plurality of recognition data is saved in an recognition server, wherein the recognition data larger than the matching data. When the mobile device firstly generates the recognition result, the recognition result is displayed for user searching reference. If the mobile device receives an recognition result from the recognition server before the mobile device completing the recognition, the recognition result replied form the recognition server is displayed on the display monitor.