LEAR-INRIA submission for the THUMOS workshop

By Heng Wang. 2013

How researchers see this paper

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

1. Zuxuan Wu
2. Yu-Gang Jiang
3. X. Wang
4. Wei David Liu
5. Kaiming He
6. Abhinav Kumar Gupta
7. Yi Yang
8. Xiangyang Xue

Citations in the corpus

• Bag of visual words and fusion methods for action recognition: Comprehensive study and good practice (2014, neutral): “Typical fusion methods include descriptor level fusion [25,54], representation level fusion [48,46], and score level fusion [41,30].”
• Temporal Segment Networks for Action Recognition in Videos (2017, neutral): “9% iDT+FV [69] 63 .”
• Temporal Segment Networks: Towards Good Practices for Deep Action Recognition (2016, unknown): “2% iDT+FV [38] 85 .”
• Hierarchical Clustering Multi-Task Learning for Joint Human Action Grouping and Recognition (2017, methodology): “From Table 5, HC-MTL can outperform the competing methods [26,47-51] and STIP+BoW.”
• Fusing Multi-Stream Deep Networks for Video Classification (2015, background): “On UCF-101, many works with competitive results are based on the hand-engineered dense trajectory features [45, 25], while our approach fully relies on the deep networks.”
• Two-Stream Convolutional Networks for Action Recognition in Videos (2014, methodology): “The learnt representation, finetuned on the UCF-101 dataset, turned out to be 20% less accurate than hand-crafted state-of-the-art trajectory-based representation [20, 27].”