Weakly Supervised Action Detection

By Parthipan Siva, Tao A. Xiang. 2011

51 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

1. G. Wang
2. Jun Liu
3. Luc Van Gool
4. Cordelia Schmid
5. Ming-Hsuan Yang
6. Dacheng Tao
7. Xiaogang Wang
8. Yi Yang

Citations in the corpus

• Skeleton-Based Online Action Prediction Using Scale Selection Network (2019, background): “action detection approaches are developed for handling offline mode that conducts detection after observing the whole long sequence [22], [23], [68].”
• Online Human Action Detection using Joint Classification-Regression Recurrent Neural Networks (2016, methodology): “To localize the action, most of previous works employ a sliding window design [8,5,9,10], which divides the sequence into overlapped clips before action recognition/classification is performed on each clip.”
• Towards Weakly-Supervised Action Localization (2016, methodology): “For the case of spatio-temporal localization, Siva and Xiang [29] define cuboids of different time lengths around detected humans, describe them with STIPs [30] and then use Multiple Instance Learning (MIL).”
• Spot On: Action Localization from Pointly-Supervised Proposals (2016, methodology): “For action localization, only the video label is used by [45,46], whereas [47] use no labels.”
• SSNet: Scale Selection Network for Online 3D Action Prediction (2018, methodology): “Several methods attempted online detection [38], while most action detection approaches are developed for handling offline mode which conducts detection after observing the whole long sequence [48, 35, 56].”
• Human Action Localization with Sparse Spatial Supervision (2017, background): “Siva and Xiang [36] apply multiple instance learning (MIL) on cuboids of various time lengths around detected humans, described by STIPs [37].”
• Fast action proposals for human action detection and search (2015, methodology): “The sequences in MSRII dataset are split into two sets (one for training and the other for testing) for [11] while we perform the standard cross-dataset evaluation (training on the KTH dataset but testing on the MSRII sequences).”
• W-TALC: Weakly-supervised Temporal Activity Localization and Classification (2018, methodology): “Person detectors are also used in [48,61] to generate person tubes, which is used to learn different Multiple Instance Learning based classifiers.”
• A flexible model for training action localization with varying levels of supervision (2018, background): “Other works assume that clip level annotations are available [7, 39].”