From Actemes to Action: A Strongly-Supervised Representation for Detailed Action Understanding

By Weiyu Zhang, Menglong Zhu, Konstantinos G. Derpanis. 2013

406 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

2D/3D Pose Estimation and Action Recognition Using Multitask Deep Learning (2018, methodology): “Penn Action .”
Jointly Modeling Deep Video and Compositional Text to Bridge Vision and Language in a Unified Framework (2015, background): “…spent decades of efforts in video content understanding, including action classification (Wang et al. 2011; Sadanand and Corso 2012; Karpathy et al. 2014), detection (Tian, Sukthankar , and Shah 2013; Zhang, Zhu, and Derpanis 2013) and tagging (Yao et al. 2013; Moxley, Mei, and Manjunath 2010).”
Learning to Generate Long-term Future via Hierarchical Prediction (2017, methodology): “In this section, we present experiments on pixel-level video prediction of human actions on the Penn Action (Weiyu Zhang & Derpanis, 2013) and Human 3.6M datasets (Ionescu et al., 2014).”
Temporal Segment Networks for Action Recognition in Videos (2017, background): “mid-level representations have been proposed for action recognition [3], [23], [47], [48], [49], [50], [51].”
Pose for Action - Action for Pose (2016, methodology): “C V ] 1 0 Fe b 20 17 Compared to [15], the pose estimation accuracy is improved by over 30%.”
Body Joint Guided 3-D Deep Convolutional Descriptors for Action Recognition (2017, methodology): “We evaluate our method on three public action datasets: 1) subJHMDB [36]; 2) Penn Action [37]; and 3) UCF101 dataset [38].”
Spot On: Action Localization from Pointly-Supervised Proposals (2016, background): “UCF-Sports [22], HOHA [23] and MSR-II [24] have up to a few hundred actions, while UCF101 [25], Penn-Action [26], and J-HMBD [27] have 1–3 thousand action clips and 3 to 24 action classes.”
PoseTrack: A Benchmark for Human Pose Estimation and Tracking (2017, methodology): “Penn Action [51] 159,633 sports (15 act.”
The Pose Knows: Video Forecasting by Generating Pose Futures (2017, methodology): “We evaluate our model on UCF-101 [41] as well as the Penn Action Dataset [56] in both pose space and video”
Temporal Cycle-Consistency Learning (2019, methodology): “We report the results on the Pouring and Penn Action datasets in Table 4.”
Learning 3D Human Dynamics From Video (2018, methodology): “Penn Action consists of 15 sports actions, with 1257 training videos and 1068 test.”
A Review on Deep Learning Techniques for Video Prediction (2020, methodology): “[54] 2017 [208] vED [115], [116] RGB,Po RGB X `2, CE,KL,AL R ? ? ? X 7”
VIBE: Video Inference for Human Body Pose and Shape Estimation (2019, methodology): “Although, we experimented with T = [8 , 16 , 32 , 64 , 128] , we chose T = 16 , as it yields the best results.”
Recognizing Human Actions as the Evolution of Pose Estimation Maps (2018, methodology): “In Table 4, poselet detected by Action Bank [54] achieves accuracy of 83.”
Deep Learning-based Human Pose Estimation: A Survey (2020, methodology): “We refer the readers to the original references for details about other datasets including LSP [90], FLIC [213], AIC-HKD [257], CrowdPose [114], Penn Action [291], J-HMDB [83], and HiEve [133].”
Chained Predictions Using Convolutional Neural Networks (2016, methodology): “For the task of pose estimation in video we use the Penn Action dataset [30], which consists of 2326 video sequences of people performing various sports.”
A Variational U-Net for Conditional Appearance and Shape Generation (2018, background): “Examples can be found at https://compvis.github.io/vunet, demonstrating the transfer of appearances from COCO to poses obtained from a video dataset [45]. the conditional image, the pose is inferred from the target image.”
Sparseness Meets Deepness: 3D Human Pose Estimation from Monocular Video (2015, methodology): “The PennAction dataset [53] is a recently introduced inthe-wild human action dataset containing 2326 challenging consumer videos.”
Joint action recognition and pose estimation from video (2015, methodology): “With only finelevel features, the performance is not very good, but when coarse/mid-level features are added in the performance is improved due to the low resolution and heavy occlusion that Method Accuracy STIP[35] 82.”
Monocular human pose estimation: A survey of deep learning-based methods (2020, methodology): “Penn Action Dataset (Zhang et al., 2013) consists of 2 , 326 videos downloaded from YouTube covering 15 actions: base-ball pitch, baseball swing, bench press, bowling, clean and jerk, golf swing, jump rope, jumping jacks, pull up, push up, sit up, squat, strum guitar, tennis forehand, and tennis…”