Learning End-to-end Video Classification with Rank-Pooling

By Basura Fernando, Stephen Gould. 2016

96 citations recorded by the source corpus.

How researchers see this paper

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

1. Jun Liu
2. Pichao Wang
3. Wanqing Li
4. Hossein Rahmani
5. Dacheng Tao
6. G. Wang
7. Amir Shahroudy
8. Yi Yang

Citations in the corpus

• On Differentiating Parameterized Argmin and Argmax Problems with Application to Bi-level Optimization (2016, methodology): “…vision where they have been applied to parameter and hyper-parameter learning [Do et al., 2007, Domke, 2012, Klatzer and Pock, 2015], image denoising [Samuel and Tappen, 2009, Ochs et al., 2015], and most recently, video activity recognition [Fernando and Gould, 2016, Fernando et al., 2016].”
• RGB-D-based Human Motion Recognition with Deep Learning: A Survey (2017, neutral): “The end-to-end learning methods with rank pooling has also been proposed in [6, 35].”
• Moments in Time Dataset: One Million Videos for Event Understanding (2018, methodology): “Dynamic images represent a video as a ranking function of its frames using rankSVM [13].”
• Rank Pooling for Action Recognition (2015, unknown): “An end-to-end CNN video classification network with rank-pooling and bi-level optimization is presented in [10].”
• Learning Action Recognition Model from Depth and Skeleton Videos (2017, background): “(4) is a bilevel optimization problem and its derivative with respect to any parameters in the model can be computed by applying the chain rule [7].”
• Deep Declarative Networks (2019, unknown): “In the context of video clas-sification, Fernando and Gould [23, 24] show how to differentiate through a rank-pooling operator [25] within a deep learning model, which involves solving a support vector regression problem.”
• Deep Bilinear Learning for RGB-D Action Recognition (2018, background): “Learning time-varying information by data-driven approaches [7, 8, 35] can generalize better to unseen sequences.”
• Hallucinating IDT Descriptors and I3D Optical Flow Features for Action Recognition With CNNs (2019, background): “Such a fusion of CNNs and handcrafted representations is timeconsuming due to pre-processing, descriptor extraction, encoding and tuning parameters.”
• Video Question Answering via Hierarchical Spatio-Temporal Attention Networks (2017, background): “On the other hand, the sequential order of the frames is also important for video representation [Fernando and Gould, 2016].”
• Generalized BackPropagation, Étude De Cas: Orthogonality (2016, methodology): “The structure within the input or the output space is also explored using convolutional neural networks Chen et al. (2015); Fernando and Gould (2016); Zheng et al. (2015).”
• Spatially and Temporally Structured Global to Local Aggregation of Dynamic Depth Information for Action Recognition (2018, methodology): “Inspired by the promising performance of recently introduced rank pooling machine [29]–[31] on action recognition from RGB video, this paper proposes to adopt rank pooling method to encode depth map sequences into dynamic images.”
• UvA-DARE (Digital Academic Repository) Self-Supervised Video Representation Learning With Odd-One-Out Networks (unknown): “Recently, supervised convolutional features obtain state-of-the-art performance either using very large video collections or using 3D convolutions [39] or by fine-tuning ImageNet pre-trained models [5, 16, 30, 43, 46].”