基于增強(qiáng)群跟蹤器和深度學(xué)習(xí)的目標(biāo)跟蹤
doi: 10.11999/JEIT141362
-
1.
(長(zhǎng)春理工大學(xué)電子信息工程學(xué)院 長(zhǎng)春 130022) ②(中國科學(xué)院長(zhǎng)春光學(xué)精密機(jī)械與物理研究所 長(zhǎng)春 130000) ③(東北師范大學(xué)計(jì)算機(jī)科學(xué)與信息技術(shù)學(xué)院 長(zhǎng)春 130117)
基金項(xiàng)目:
國家自然科學(xué)基金(61172111)和吉林省科技廳項(xiàng)目(20090512, 20100312)資助課題
Target Tracking Based on Enhanced Flock of Tracker and Deep Learning
-
1.
(School of Electronic Information Engineering, Changchun University of Science and Technology, Changchun 130022, China)
-
摘要: 為解決基于外觀模型和傳統(tǒng)機(jī)器學(xué)習(xí)目標(biāo)跟蹤易出現(xiàn)目標(biāo)漂移甚至跟蹤失敗的問題，該文提出以跟蹤-學(xué)習(xí)-檢測(cè)(TLD)算法為框架，基于增強(qiáng)群跟蹤器(FoT)和深度學(xué)習(xí)的目標(biāo)跟蹤算法。FoT實(shí)現(xiàn)目標(biāo)的預(yù)測(cè)與跟蹤，增添基于時(shí)空上下文級(jí)聯(lián)預(yù)測(cè)器提高預(yù)測(cè)局部跟蹤器的成功率，快速隨機(jī)采樣一致性算法評(píng)估全局運(yùn)動(dòng)模型，提高目標(biāo)跟蹤的精確度。深度去噪自編碼器和支持向量機(jī)分類器構(gòu)建深度檢測(cè)器，結(jié)合全局多尺度掃描窗口搜索策略檢測(cè)可能的目標(biāo)。加權(quán)P-N學(xué)習(xí)對(duì)樣本加權(quán)處理，提高分類器的分類精確度。與其它跟蹤算法相比較，在復(fù)雜環(huán)境下，不同圖片序列實(shí)驗(yàn)結(jié)果表明，該算法在遮擋、相似背景等條件下具有更高的準(zhǔn)確度和魯棒性。
-
關(guān)鍵詞:
- 計(jì)算機(jī)視覺 /
- 群跟蹤器 /
- 跟蹤-學(xué)習(xí)-檢測(cè) /
- 深度學(xué)習(xí) /
- 支持向量機(jī) /
- 深度檢測(cè)器
Abstract: To solve the problem that the tracking algorithm often leads to drift and failure based on the appearance model and traditional machine learning, a tracking algorithm is proposed based on the enhanced Flock of Tracker (FoT) and deep learning under the Tracking-Learning-Detection (TLD) framework. The target is predicted and tracked by the FoT, the cascaded predictor is added to improve the precision of the local tracker based on the spatio-temporal context, and the global motion model is evaluated by the speed-up random sample consensus algorithm to improve the accuracy. A deep detector is composed of the stacked denoising autoencoder and Support Vector Machine (SVM), combines with a multi-scale scanning window with global search strategy to detect the possible targets. Each sample is weighted by the weighted P-N learning to improve the precision of the deep detector. Compared with the state-of-the-art trackers, according to the results of experiments on variant challenging image sequences in the complex environment, the proposed algorithm has more accuracy and better robust, especially for the occlusions, the background clutter and so on. -
Wu Y, Lim J, and Yang M H. Online object tracking: A benchmark[C]. Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, Portland, USA, 2013: 2411-2418. Ross D A, Lim J, Lin R S, et al.. Incremental learning for robust visual tracking[J]. International Journal of Computer Vision, 2008, 77(3): 125-141. Babenko B, Yang M H, and Belongie S. Robust object tracking with online multiple instance learning[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011, 33(8): 1619-1632. 陳東成, 朱明, 高文, 等. 在線加權(quán)多示例學(xué)習(xí)實(shí)時(shí)目標(biāo)跟蹤[J]. 光學(xué)精密工程, 2014, 22(6): 1661-1667. Chen Dong-cheng, Zhu Ming, Gao Wen, et al.. Real-time object tracking via online weighted multiple instance learning [J]. Optics and Precision Engineerin, 2014, 22(6): 1661-1667. He S F, Yang Q X, Rynson L, et al.. Visual Tracking via Locality Sensitive Histograms[C]. Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, Portland, USA, 2013: 2427-2434. Grabner H, Grabner M, and Bischof H. Real-time tracking via online boosting[C]. Proceedings of British Machine Vision Conference, Edinburgh, UK, 2006: 47-56. Grabner H, Leistner C, and Bischof H. Semi-supervised on-line boosting for robust tracking[C]. Proceedings of European Conference on Computer Vision, Berlin, Germany, 2008: 234-247. 顏佳, 吳敏淵. 遮擋環(huán)境下采用在線Boosting的目標(biāo)跟蹤[J]. 光學(xué)精密工程, 2012, 20(2): 439-446. Yan Jia and Wu Ming-yuan. On-line boosting based target tracking under occlusion[J]. Optics and Precision Engineering, 2012, 20(2): 439-446. Kalal Z, Matas J, and Mikolajczyk K. P-N learning: bootstrapping binary classifiers by structural constraints[C]. Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, New York, USA, 2010: 49-56. 鄭胤, 陳權(quán)崎, 章毓晉. 深度學(xué)習(xí)及其在目標(biāo)和行為識(shí)別中的新進(jìn)展[J]. 中國圖像圖形學(xué)報(bào), 2014, 19(2): 175-184. Zheng Ying, Chen Quan-qi, and Zhang Yu-jin. Deep learning and its new progress in object and behavior recognition[J]. Journal of Image and Graphic, 2014, 19(2): 175-184. Tomas V and Jiri M. Robustifying the flock of trackers[C]. Proceedings of Computer Vision Winter Workshop, Graz, Austria, 2011: 91-97. 周鑫, 錢秋朦, 葉永強(qiáng), 等. 改進(jìn)后的TLD視頻目標(biāo)跟蹤方法[J]. 中國圖象圖形學(xué)報(bào), 2013, 18(9): 1115-1123. Zhou Xin, Qian Qiu-meng, Ye Yong-qiang, et al.. Improved TLD visual target tracking algorithm[J]. Journal of Image and Graphic, 2013, 18(9): 1115-1123. Kalal Z, Mikolajczyk K, and Matas J. Tracking-learning- detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2012, 34(7): 1409-1422. Zhang K, Zhang L, Liu Q, et al.. Fast visual tracking via dense spatio-temporal context learning[C]. Proceedings of European Conference on Computer Vision, Zurich, Switzerland, 2014: 127-141. Botterill T, Mills S, and Green R D. New conditional sampling strategies for speeded-up RANSAC[C]. Proceedings of British Machine Vision Conference, London, UK, 2009: 1-11. Vincent P, Larochelle H, Lajoie I, et al.. Stacked denoising autoencoders: learning useful representations in a deep network with a local denoising criterion[J]. Journal of Machine Learning Research, 2010, 11(2): 3371-3408. Tang Yi-chuan. Deep learning using linear support vector machines[C]. Proceedings of International Conference on Machine Learning: Challenges in Representational Learning Workshop, Atlanta, USA, 2013: 266-272. Hinton G E and Salakhutdinov R R. Reducing the dimensionality of data with neural networks[J]. Science, 2006, 313(5786): 504-507. Torralba A, Fergus R, and Freeman W T. 80 million tiny images: a large data set for nonparametric object and scene recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2008, 30(11): 1958-1970. 高文, 湯洋, 朱明. 復(fù)雜背景下目標(biāo)檢測(cè)的級(jí)聯(lián)分類器算法研究[J]. 物理學(xué)報(bào), 2014, 63(9): 094204. Gao Wen, Tang Yang, and Zhu Ming. Study on the cascade classifier in target detection under complex background[J]. Acta Physica Sinica, 2014, 63(9): 094204. Collins R T, Zhou X H, and Teh S K. An open source tracking test bed and evaluation web site[C]. Proceedings of IEEE International Workshop on Performance Evaluation of Tracking and Surveillance, Breckenridge, USA, 2005: 17-24. Stalder S, Grabner H, and Van G L. Beyond semi-supervised tracking: tracking should be as simple as detection, but not simpler than recognition[C]. Proceedings of IEEE Conference on Computer Vision Workshops, Kyoto, Japan, 2009: 1409-1416. Dinh T B, Vo N, and Medion G. Context tracker: exploring supporters and distracters in unconstrained environments[C]. Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, Providence, USA, 2011: 1177-1184. Qian Yu, Thang B D, and Gerard M. Online tracking and reacquisition using co-trained generative and discriminative trackers[C]. Proceedings of European Conference on Computer Vision, Marseille, France, 2008: 678-691. -
計(jì)量
- 文章訪問數(shù): 1340
- HTML全文瀏覽量: 117
- PDF下載量: 1577
- 被引次數(shù): 0
下載:
