A Pursuit of Temporal Accuracy in General Activity Detection
arXiv:1703.02716
Abstract
Detecting activities in untrimmed videos is an important but challenging task. The performance of existing methods remains unsatisfactory, e.g., they often meet difficulties in locating the beginning and end of a long complex action. In this paper, we propose a generic framework that can accurately detect a wide variety of activities from untrimmed videos. Our first contribution is a novel proposal scheme that can efficiently generate candidates with accurate temporal boundaries. The other contribution is a cascaded classification pipeline that explicitly distinguishes between relevance and completeness of a candidate instance. On two challenging temporal activity detection datasets, THUMOS14 and ActivityNet, the proposed framework significantly outperforms the existing state-of-the-art methods, demonstrating superior accuracy and strong adaptivity in handling activities with various temporal structures.
References in corpus (9)
- UCF101: A Dataset of 101 Human Actions Classes From Videos in The Wild
- Caffe: Convolutional Architecture for Fast Feature Embedding
- Action Recognition with Trajectory-Pooled Deep-Convolutional Descriptors
- Rethinking the Inception Architecture for Computer Vision
- Rich feature hierarchies for accurate object detection and semantic segmentation
- Temporal Action Localization in Untrimmed Videos via Multi-stage CNNs
- Learning Spatiotemporal Features with 3D Convolutional Networks
- Untrimmed Video Classification for Activity Detection: submission to ActivityNet Challenge
- Temporal Activity Detection in Untrimmed Videos with Recurrent Neural Networks
Cited by in corpus (37)
- Revisiting Anchor Mechanisms for Temporal Action Localization
- BSN: Boundary Sensitive Network for Temporal Action Proposal Generation
- BMN: Boundary-Matching Network for Temporal Action Proposal Generation
- Temporal Convolution Based Action Proposal: Submission to ActivityNet 2017
- Weakly Supervised Action Localization by Sparse Temporal Pooling Network
- Look Closer to Ground Better: Weakly-Supervised Temporal Grounding of Sentence in Video
- Step-by-step Erasion, One-by-one Collection: A Weakly Supervised Temporal Action Detector
- BSN++: Complementary Boundary Regressor with Scale-Balanced Relation Modeling for Temporal Action Proposal Generation
- Multi-granularity Generator for Temporal Action Proposal
- CTAP: Complementary Temporal Action Proposal Generation
- YH Technologies at ActivityNet Challenge 2018
- Weakly Supervised Temporal Adjacent Network for Language Grounding
- Segregated Temporal Assembly Recurrent Networks for Weakly Supervised Multiple Action Detection
- Bottom-Up Temporal Action Localization with Mutual Regularization
- Boundary Content Graph Neural Network for Temporal Action Proposal Generation
- Self-Supervised Learning for Semi-Supervised Temporal Action Proposal
- Deep Concept-wise Temporal Convolutional Networks for Action Localization
- Action Unit Memory Network for Weakly Supervised Temporal Action Localization
- Action Completion: A Temporal Model for Moment Detection
- Diagnosing Error in Temporal Action Detectors
- A Self-Adaptive Proposal Model for Temporal Action Detection based on Reinforcement Learning
- Learning to Localize Actions from Moments
- Scale Matters: Temporal Scale Aggregation Network for Precise Action Localization in Untrimmed Videos
- CMSN: Continuous Multi-stage Network and Variable Margin Cosine Loss for Temporal Action Proposal Generation
- Adversarial Seeded Sequence Growing for Weakly-Supervised Temporal Action Localization
- Class Semantics-based Attention for Action Detection
- Cascaded Pyramid Mining Network for Weakly Supervised Temporal Action Localization
- Multi-Level Temporal Pyramid Network for Action Detection
- Temporal Action Localization using Long Short-Term Dependency
- ZSTAD: Zero-Shot Temporal Activity Detection
- Gaussian Temporal Awareness Networks for Action Localization
- Relation-Aware Pyramid Network (RapNet) for temporal action proposal
- Multi-Granularity Fusion Network for Proposal and Activity Localization: Submission to ActivityNet Challenge 2019 Task 1 and Task 2
- Temporal Action Detection by Joint Identification-Verification
- Complementary Boundary Generator with Scale-Invariant Relation Modeling for Temporal Action Localization: Submission to ActivityNet Challenge 2020
- TruNet: Short Videos Generation from Long Videos via Story-Preserving Truncation
- Accurate Temporal Action Proposal Generation with Relation-Aware Pyramid Network