Video Event Detection by Inferring Temporal Instance Labels

By Kuan-Ting Lai, Felix X. Yu, Ming-Syan Chen, Shih-Fu Chang. 2014

133 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

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

Citations in the corpus

WILDCAT: Weakly Supervised Learning of Deep ConvNets for Image Classification, Pointwise Localization and Segmentation (2017, unknown): “Recent alternatives include Global Average Pooling (GAP) [70], soft max in LSE pooling [58], Learning from Label Proportion (LLP) [65, 36], and top max scoring [39].”
Revealing Event Saliency in Unconstrained Video Collection (2017, unknown): “Similarly, Lai et al. [15] also represented each video as a collection of instances and their goal is to learn an instance-level event detection model based on the video-level labels.”
Multiple instance learning: A survey of problem characteristics and applications (2016, methodology): “It has been used to recognize complex events such as “attempting a board trick” or “birthday party” [8, 113].”
Weakly supervised learning via statistical sufficiency (2016, background): “The problem of detecting events in videos by classifying singular frames is characterized by the same kind of weak supervision [Lai et al., 2014].”
(Almost) No Label No Cry (2014, unknown): “The degree of violation is measured as ‖B± − B±‖F , where B± is the homogeneity assumption matrix, that replaces all bσj by bσ for σ ∈ {−1, 1}, see eq.”
Temporal Action Localization in Untrimmed Videos via Multi-stage CNNs (2016, unknown): “When training data only have video-level category labels but no temporal annotations, researchers formulated this as weakly supervised problems or multiple instance learning problems to learn the key evidences in untrimmed videos and temporally localize actions by selecting key instances [22, 23].”
DevNet: A Deep Event Network for multimedia event detection and evidence recounting (2015, background): “To overcome these limitations, [18] and [19] formulated the problem as a multipleinstance learning problem, aiming at learning an instancelevel event detection and recounting model by selecting the informative shots or key fames during the training process.”
WELDON: Weakly Supervised Learning of Deep Convolutional Neural Networks (2016, methodology): “In [24], the LLP method of [51] is explicitly applied to MIL problems, in the context of video event detection.”
Fusing Multi-Stream Deep Networks for Video Classification (2015, background): “Compared with human actions, the temporal clues of these classes are more obscure and thus difficult to be captured.”