Semantic event detection using cross-domain knowledge

1. A method for facilitating semantic event classification of a group of image records related to an event, the method using an event detector system for providing:

extracting a plurality of visual features from each of the image records;

wherein the visual features include segmenting an image record into a number of regions, in which the visual features are extracted;

generating a plurality of concept scores for each of the image records using the visual features, wherein each concept score corresponds to a visual concept and each concept score is indicative of a probability that the image record includes the visual concept;

determining a pair-wise similarity between pairs of data points corresponding to training events to produce a codebook of semantic events by applying spectral clustering to group the data points into different clusters, with each cluster corresponding to one code word, based on the determined pair-wise similarities;

mapping the training events to the codebook of semantic events to generate a first feature vector corresponding to each training event;

training an event classifier based on the first feature vectors corresponding to the training events;

generating a second feature vector by calculating a pairwise similarity between the concept scores of the image records and concept scores for predetermined training data points;

mapping the second feature vectors to the codebook of semantic events to produce a third feature vector; and

supplying the third feature vector to the event classifier that identifies at least one semantic event.

2. The method for facilitating semantic event classification as claimed in claim 1 , further comprising using cross-domain learning to generate the feature vector.

3. The method for facilitating semantic event classification as claimed in claim 2 , wherein the cross-domain learning is based on image-level or region-level features.

4. The method for facilitating semantic event classification as claimed in claim 1 , wherein the image records include at least one digital still image and at least one video segment.

5. The method for facilitating semantic event classification as claimed in claim 4 , wherein the extracting of a plurality of visual features includes extracting a keyframe from the video segment and extracting the plurality of visual features from both the keyframe and the digital still image.

6. The method for facilitating semantic event classification as claimed in claim 5 , wherein the generation of the concept scores includes generating initial concept scores for each keyframe and each digital still image corresponding to each of the extracted visual features.

7. The method for facilitating semantic event classification as claimed in claim 6 , wherein the generation of the concept scores further includes generating ensemble concept scores for each keyframe and each digital still image based on the initial concept scores.

8. A method for facilitating semantic event classification as claimed in claim 7 , wherein the ensemble concept scores are generated by fusing the initial concept scores for each extracted visual feature for a given keyframe or a given digital still image.

9. A method for facilitating semantic event classification as claimed in claim 1 , further comprising tagging each of the image records with the semantic event classifier.