Action Recognition in Movie Scenes using Deep Features of Keyframes
2
조회
0
다운로드
7
인용
0
참고문헌
키워드
초록
Recently, researchers of computer vision have focused on human action recognition in video clips and have used it for applications in various domains such as surveillance and sports. In this paper, we have recognized human action in movies clips using deep features of keyframes. Firstly, k-mean clustering is used to achieve representative frames (keyframes) from action videos. Clustering removes redundant frames from action videos, thereby reducing the computational complexity of the algorithm. Secondly, we fine-tuned convolutional neural network (CNN) model called Alexnet for 12 movies actions. Finally, we feed those representative frames of action video clips to the fine-tuned CNN classifier. The proposed algorithm is experimentally tested on Holywood2 dataset and the obtained results are better in terms of accuracy compared to state-of-the-art hand crafted features extraction based action recognition methods.
Abstract
Recently, researchers of computer vision have focused on human action recognition in video clips and have used it for applications in various domains such as surveillance and sports. In this paper, we have recognized human action in movies clips using deep features of keyframes. Firstly, k-mean clustering is used to achieve representative frames (keyframes) from action videos. Clustering removes redundant frames from action videos, thereby reducing the computational complexity of the algorithm. Secondly, we fine-tuned convolutional neural network (CNN) model called Alexnet for 12 movies actions. Finally, we feed those representative frames of action video clips to the fine-tuned CNN classifier. The proposed algorithm is experimentally tested on Holywood2 dataset and the obtained results are better in terms of accuracy compared to state-of-the-art hand crafted features extraction based action recognition methods.
참고문헌 (15)
- Ramanathan, M., "Human action recognition with video data:research and evaluation challenges", IEEE Transactions on Human-Machine Systems, 44(5), pp.650~663 (2014)
- Youssef, M., "Human action recognition using hull convexity defect features with multi-modality setups", Pattern Recognition Letters, 34(15), pp.1971~1979 (2013)
- Aggarwal, J.K., "Human motion analysis: A review", in Nonrigid and Articulated Motion Workshop, 1997. Proceedings., IEEE (1997), IEEE
- Klaser, A., "A spatio-temporal descriptor based on 3d-gradients", in BMVC 2008-19th British Machine Vision Conference (2008), British Machine Vision Association
- Wang, H., "Dense trajectories and motion boundary descriptors for action recognition", International journal of computer vision, 103(1), pp.60~79 (2013)
- Kviatkovsky, I., "Online action recognition using covariance of shape and motion", Computer Vision and Image Understanding, 129, pp.15~26 (2014)
- Chen, L., "ReadingAct RGB-D action dataset and human action recognition from local features", Pattern Recognition Letters, 50, pp.159~169 (2014)
- Laptev, I., "Learning realistic human actions from movies", in Computer Vision and Pattern Recognition, 2008. CVPR 2008. IEEE Conference on (2008), IEEE
- Amin ullah, "M.S. Content based Image Retrieval using Local Shape and Color Features in Lab* Color Space", in The 2nd International Conference on Next Generation Computing 2017 (2017), Korean Institute of Next Generation Computing
- Sun, L., "DL-SFA: deeply-learned slow feature analysis for action recognition", in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (2014)
- Du, Y., "Hierarchical recurrent neural network for skeleton based action recognition", in Proceedings of the IEEE conference on computer vision and pattern recognition (2015)
- Hartigan, J.A., "Algorithm AS 136: A k-means clustering algorithm", Journal of the Royal Statistical Society. Series C (Applied Statistics), 28(1), pp.100~108 (1979)
- Krizhevsky, A., "Imagenet classification with deep convolutional neural networks", Advances in neural information processing systems (2012)
- 울라 아민, "Analyzing Pedestrian Parts using Deep Features for Person Re-Identification", 한국차세대컴퓨팅학회 논문지, 13(2), pp.82~92 (2017), 한국차세대컴퓨팅학회
- Ejaz, N., "Efficient visual attention based framework for extracting key frames from videos", Signal Processing: Image Communication, 28(1), pp.34~44 (2013)