DeepAct: A Deep Neural Network Model for Activity Detection in Untrimmed Videos
1
조회
0
다운로드
1
인용
0
참고문헌
키워드
초록
We propose a novel deep neural network model for detecting human activities in untrimmed videos. Theprocess of human activity detection in a video involves two steps: a step to extract features that are effective inrecognizing human activities in a long untrimmed video, followed by a step to detect human activities fromthose extracted features. To extract the rich features from video segments that could express unique patternsfor each activity, we employ two different convolutional neural network models, C3D and I-ResNet. Fordetecting human activities from the sequence of extracted feature vectors, we use BLSTM, a bi-directionalrecurrent neural network model. By conducting experiments with ActivityNet 200, a large-scale benchmarkdataset, we show the high performance of the proposed DeepAct model.
Abstract
We propose a novel deep neural network model for detecting human activities in untrimmed videos. Theprocess of human activity detection in a video involves two steps: a step to extract features that are effective inrecognizing human activities in a long untrimmed video, followed by a step to detect human activities fromthose extracted features. To extract the rich features from video segments that could express unique patternsfor each activity, we employ two different convolutional neural network models, C3D and I-ResNet. Fordetecting human activities from the sequence of extracted feature vectors, we use BLSTM, a bi-directionalrecurrent neural network model. By conducting experiments with ActivityNet 200, a large-scale benchmarkdataset, we show the high performance of the proposed DeepAct model.
참고문헌 (19)
- H. Wang, "Action recognition with improved trajectories", Proceedings of IEEE International Conference on Computer Vision (ICCV-13), pp.3551~3558 (2013)
- L. Wang, "Video action detection with relational dynamic-poselets", Proceedings of European Conference on Computer Vision (ECCV-14), pp.565~580 (2014)
- M. Schuster, "Bidirectional recurrent neural networks", IEEE Transactions on Signal Processing, 45(11), pp.2673~2681 (1997)[DOI]
- F. C. Heilbron, "ActivityNet: a large-scale video benchmark for human activity understanding", Proceedings of IEEE International Conference on Computer Vision and Pattern Recognition (CVPR-15), pp.961~970 (2015)
- S. Ji, "3D convolutional neural networks for human action recognition", IEEE Transactions on Pattern Analysis and Machine Intelligence, 35(1), pp.221~231 (2013)[DOI]
- K. Simonyan, "Two-stream convolutional networks for action recognition in videos", Proceedings of the International Conference on Neural Information Processing Systems (NIPS-14), pp.568~576 (2014)
- J. Zheng, "Cross-view action recognition via transferable dictionary learning", IEEE Transactions on Image Processing, 5(6), pp.2542~2556 (2016)
- R. Wang, "UTS at ActivityNet 2016", ActivityNet Large Scale Activity Recognition Challenge Workshop, pp.1~6 (2016)
- Z. Shou, "CDC: convolutional-de-convolutional networks for precise temporal action localization in untrimmed videos", Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp.5734~5743 (2017)
- "A pursuit of temporal accuracy in general activity detection" (2017)[URL]
- L. Wang, "Temporal segment networks: towards good practices for deep action recognition", Proceedings of European Conference on Computer Vision (ECCV), pp.20~36 (2016)
- "Untrimmed video classification for activity detection: submission to ActivityNet challenge" (2016)[URL]
- S. Karaman, "Fast saliency based pooling of fisher encoded dense trajectories", Proceedings of European Conference on Computer Vision (ECCV) Workshop, pp.1~4 (2014)
- Z. Shou, "Temporal action localization in untrimmed videos via multi-stage CNNs", Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp.1049~1058 (2016)
- L. Wang, "Action recognition and detection by combining motion and appearance features", Proceedings of European Conference on Computer Vision (ECCV), pp.1~6 (2014)
- D. Tran, "Learning spatiotemporal features with 3D convolutional networks", Proceedings of IEEE International Conference on Computer Vision (ICCV), pp.4489~4497 (2015)
- V. Escorcia, "DAPs: deep action proposals for action understanding", Proceedings of European Conference on Computer Vision (ECCV), pp.768~784 (2016)
- "Temporal activity detection in untrimmed videos with recurrent neural networks" (2017)[URL]
- K. He, "Deep residual learning for image recognition", Proceedings of IEEE International Conference on Computer Vision and Pattern Recognition (CVPR), pp.770~778 (2016)