Towards Scalable Summarization of Consumer Videos Via Sparse Dictionary Selection

By Yang Cong, Junsong Yuan, Jiebo Luo. IEEE transactions on multimedia, 2012

294 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Signal Processing

View Paper: https://www.semanticscholar.org/paper/c6f27e57cfcbd71aa01e2f91a314f84dd093b4bc

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Signal Processing

Citations in the corpus

Enhancing Video Summarization via Vision-Language Embedding (2017, methodology): “, using notions of sparsity, graph connectedness, or statistical significance) [2, 17, 49] or tried to learn them using implicit or explicit supervision [30, 35, 43, 47].”
Deep Attentive Video Summarization With Distribution Consistency Learning (2020, methodology): “It splits the neurons of a regular LSTM [24] into two directions: one for positive time direction (forward states) and the other for negative time direction (backward states), which”
Category-Specific Video Summarization (2014, unknown): “In [16,17], the main task is to remove redundant video footage, which is detected as easy to reconstruct based on sparse coding from the rest of the video.”
A Weighted Sparse Sampling and Smoothing Frame Transition Approach for Semantic Fast-Forward First-Person Videos (2018, methodology): “[3] formulate the problem of video summarization as a dictionary selection problem.”
TVSum: Summarizing web videos using titles (2015, background): “, based on content frequency [52] or non-redundancy [12, 50]; while simple and efficient, visual frequency is not directly related to topicality.”
Video Summarization With Attention-Based Encoder–Decoder Networks (2017, methodology): “ssary information in a sequence [38], we select it as an encoder for taking the temporal relation of video frames into consideration. The principle of BiLSTM is to split the neurons of a regular LSTM [30] into two directions, one for positive time direction (forward states), and the other for negati”
Generating Diverse Image Datasets with Limited Labeling (2016, background): “, l2,1 regularizer is to induce row level sparsity in representative selection [8, 11], whereas the second constraint tries to select images that are less correlated with images in current dataset.”