Abstract: A multimodal and real-time method for filtering sensitive content, receiving as input a digital video stream, the method including segmenting digital video into video fragments along the video timeline; extracting features containing significant information from the digital video input on sensitive media; reducing the semantic difference between each of the low-level video features, and the high-level sensitive concept; classifying the video fragments, generating a high-level label (positive or negative), with a confidence score for each fragment representation; performing high-level fusion to properly match the possible high-level labels and confidence scores for each fragment; and predicting the sensitive time by combining the labels of the fragments along the video timeline, indicating the moments when the content becomes sensitive.

Abstract: The proposed method is used for classification in open-set scenarios, wherein often it is not possible to first obtain the training data for all possible classes that may arise during the testing stage. During the test phase, test samples belonging to one of the classes used in the training phase are classified based on a ratio between similarity scores, as known correct class and test samples belonging to any other class are to be rejected and classified as unknown.

Abstract: A multimodal and real-time method for filtering sensitive content, receiving as input a digital video stream, the method including segmenting digital video into video fragments along the video timeline; extracting features containing significant information from the digital video input on sensitive media; reducing the semantic difference between each of the low-level video features, and the high-level sensitive concept; classifying the video fragments, generating a high-level label (positive or negative), with a confidence score for each fragment representation; performing high-level fusion to properly match the possible high-level labels and confidence scores for each fragment; and predicting the sensitive time by combining the labels of the fragments along the video timeline, indicating the moments when the content becomes sensitive.

Abstract: The proposed method is used for classification in open-set scenarios, wherein often it is not possible to first obtain the training data for all possible classes that may arise during the testing stage. During the test phase, test samples belonging to one of the classes used in the training phase are classified based on a ratio between similarity scores, as known correct class and test samples belonging to any other class are to be rejected and classified as unknown.