DAPs: Deep Action Proposals for Action Understanding

By Victor Escorcia, Fabian Caba Heilbron, Juan Carlos Niebles, Bernard Ghanem. 2016

426 citations recorded by the source corpus.

Subfields: Artificial Intelligence, 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 Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Towards Automatic Learning of Procedures From Web Instructional Videos (2017, methodology): “DAPs determines ﬁxed proposal locations by clustering over the ground-truth segments, while our model learns to localize procedures with anchor offsets, which is a generalization of the location pattern from training to testing instead of directly transferring.”
Dense-Captioning Events in Videos (2017, methodology): “To tackle this challenge, we design an event proposal module to be a variant of DAPs [10] that can detect longer events.”
RGB-D-based Human Motion Recognition with Deep Learning: A Survey (2017, background): “[30] introduced the Deep Action Proposals (DAPs) that generate temporal action proposals from long untrimmed videos for action detection and classification (see Figure.”
Action Genome: Actions As Compositions of Spatio-Temporal Scene Graphs (2019, background): “Such findings hint that it is possible to predict when actions begin and end, and have inspired hundreds of Computer Vision datasets, models, and algorithms to study tasks like action recognition [19, 37, 71, 80, 81, 82].”
ConvNet Architecture Search for Spatiotemporal Feature Learning (2017, background): “Some baselines in [13, 43, 25, 3] are SVM classifiers trained on a pool of features that do not specifically address the detection problem.”
Temporal Action Detection with Structured Segment Networks (2017, methodology): “Particularly, we compare our TAG scheme with common sliding windows as well as other state-of-the-art proposal methods, including SCNN-prop, a proposal networks presented in [38], TAP [7], DAP [6].”
What Do I Annotate Next? An Empirical Study of Active Learning for Action Localization (2018, background): “To unburden the computational requirements of sliding windows, a new line of work studies the use of action proposals to quickly scan a video in an attempt to reduce the search space [6,7,10,20,24,56].”
Gaussian Temporal Awareness Networks for Action Localization (2019, background): “[5] leverage Long Short-Term Memory cells to learn an appropriate encoding of a video sequence as a set of discriminative states to indicate proposal scores.”
TURN TAP: Temporal Unit Regression Network for Temporal Action Proposals (2017, methodology): “We compare TURN with the state-of-the-art methods under AR-AN, AR-F, Recall@AN-tIoU, Recall@F-tIoU metrics.”
CompILE: Compositional Imitation Learning and Execution (2018, background): “Various solutions for supervised sequence segmentation or task decomposition exist which require varying degrees of supervision (Graves, 2012; Escorcia et al., 2016; Krishna et al., 2017; Shiarlis et al., 2018).”
G-TAD: Sub-Graph Localization for Temporal Action Detection (2019, methodology): “For proposal generation, they either use fixed handcrafted anchors [5, 6, 13, 15, 37] , or adaptively form proposal candidates by connecting potential start and end frames [56, 30].”
Temporal Localization of Moments in Video Collections with Natural Language (2019, background): “While indexing only action proposals [8, 9] may be a solution to reducing the index size, such methods may discard relevant moments that a user may want to query.”
Learning Salient Boundary Feature for Anchor-free Temporal Action Localization (2021, neutral): “S-CNN [33] — 60 DAP [9] — 134 CDC [32] TITAN Xm 500 SS-TAD [4] TITAN Xm 701 R-C3D [38] TITAN Xm 569 R-C3D [38] TITAN Xp 1030 Ours 1080 Ti 3259 Ours V100 4057”
Graph Distillation for Action Detection with Privileged Modalities (2017, background): “The field of action classification [3,50,52] and action detection [2,11,14,65] in RGB videos has been studied by the computer vision community for decades.”
Rethinking the Faster R-CNN Architecture for Temporal Action Localization (2018, methodology): “Inspired by the recent success of region-based detectors in object detection [18, 17], many recent approaches adopt a two-stage, proposal-plus-classification framework [6, 36, 12, 3, 4, 35, 56], i.”
Human Action Recognition and Prediction: A Survey (2018, background): “The idea of their action proposals inspired many later research [49, 233,277,331,298,62,22].”
CDC: Convolutional-De-Convolutional Networks for Precise Temporal Action Localization in Untrimmed Videos (2017, methodology): “[9] built a temporal action proposal system based on Long-Short Term Memory (LSTM); Yeung et al.”
Cascaded Boundary Regression for Temporal Action Detection (2017, methodology): “DAPs [3] used LSTM networks to encode a video stream and produce proposals inside the video stream.”