A scale-adaptive object-tracking algorithm with occlusion detection
Abstract The methods combining correlation filters (CFs) with the features of convolutional neural network (CNN) are good at object tracking. However, the high-level features of a typical CNN without residual structure suffer from the shortage of fine-grained information, it is easily affected by si...
| Main Authors: | , , , , |
|---|---|
| 格式: | 文件 |
| 语言: | English |
| 出版: |
SpringerOpen
2020-02-01
|
| 丛编: | EURASIP Journal on Image and Video Processing |
| 主题: | |
| 在线阅读: | http://link.springer.com/article/10.1186/s13640-020-0496-6 |
| _version_ | 1828304621937885184 |
|---|---|
| author | Yue Yuan Jun Chu Lu Leng Jun Miao Byung-Gyu Kim |
| author_facet | Yue Yuan Jun Chu Lu Leng Jun Miao Byung-Gyu Kim |
| author_sort | Yue Yuan |
| collection | DOAJ |
| description | Abstract The methods combining correlation filters (CFs) with the features of convolutional neural network (CNN) are good at object tracking. However, the high-level features of a typical CNN without residual structure suffer from the shortage of fine-grained information, it is easily affected by similar objects or background noise. Meanwhile, CF-based methods usually update filters at every frame even when occlusion occurs, which degrades the capability of discriminating the target from background. A novel scale-adaptive object-tracking method is proposed in this paper. Firstly, the features are extracted from different layers of ResNet to produce response maps, and then, in order to locate the target more accurately, these response maps are fused based on AdaBoost algorithm. Secondly, to prevent the filters from updating when occlusion occurs, an update strategy with occlusion detection is proposed. Finally, a scale filter is used to estimate the target scale. The experimental results demonstrate that the proposed method performs favorably compared with several mainstream methods especially in the case of occlusion and scale change. |
| first_indexed | 2024-04-13T14:14:46Z |
| format | Article |
| id | doaj.art-ee1e19c64b974d019514a9e5207b5328 |
| institution | Directory Open Access Journal |
| issn | 1687-5281 |
| language | English |
| last_indexed | 2024-04-13T14:14:46Z |
| publishDate | 2020-02-01 |
| publisher | SpringerOpen |
| record_format | Article |
| series | EURASIP Journal on Image and Video Processing |
| spelling | doaj.art-ee1e19c64b974d019514a9e5207b53282022-12-22T02:43:42ZengSpringerOpenEURASIP Journal on Image and Video Processing1687-52812020-02-012020111510.1186/s13640-020-0496-6A scale-adaptive object-tracking algorithm with occlusion detectionYue Yuan0Jun Chu1Lu Leng2Jun Miao3Byung-Gyu Kim4Key Laboratory of Jiangxi Province for Image Processing and Pattern Recognition, Nanchang Hangkong UniversityKey Laboratory of Jiangxi Province for Image Processing and Pattern Recognition, Nanchang Hangkong UniversityKey Laboratory of Jiangxi Province for Image Processing and Pattern Recognition, Nanchang Hangkong UniversityKey Laboratory of Jiangxi Province for Image Processing and Pattern Recognition, Nanchang Hangkong UniversityIntelligent Vision Processing Lab, Department of IT Engineering, Sookmyung Women’s UniversityAbstract The methods combining correlation filters (CFs) with the features of convolutional neural network (CNN) are good at object tracking. However, the high-level features of a typical CNN without residual structure suffer from the shortage of fine-grained information, it is easily affected by similar objects or background noise. Meanwhile, CF-based methods usually update filters at every frame even when occlusion occurs, which degrades the capability of discriminating the target from background. A novel scale-adaptive object-tracking method is proposed in this paper. Firstly, the features are extracted from different layers of ResNet to produce response maps, and then, in order to locate the target more accurately, these response maps are fused based on AdaBoost algorithm. Secondly, to prevent the filters from updating when occlusion occurs, an update strategy with occlusion detection is proposed. Finally, a scale filter is used to estimate the target scale. The experimental results demonstrate that the proposed method performs favorably compared with several mainstream methods especially in the case of occlusion and scale change.http://link.springer.com/article/10.1186/s13640-020-0496-6Scale adaptionObject trackingResnetCorrelation filtersOcclusion detection |
| spellingShingle | Yue Yuan Jun Chu Lu Leng Jun Miao Byung-Gyu Kim A scale-adaptive object-tracking algorithm with occlusion detection EURASIP Journal on Image and Video Processing Scale adaption Object tracking Resnet Correlation filters Occlusion detection |
| title | A scale-adaptive object-tracking algorithm with occlusion detection |
| title_full | A scale-adaptive object-tracking algorithm with occlusion detection |
| title_fullStr | A scale-adaptive object-tracking algorithm with occlusion detection |
| title_full_unstemmed | A scale-adaptive object-tracking algorithm with occlusion detection |
| title_short | A scale-adaptive object-tracking algorithm with occlusion detection |
| title_sort | scale adaptive object tracking algorithm with occlusion detection |
| topic | Scale adaption Object tracking Resnet Correlation filters Occlusion detection |
| url | http://link.springer.com/article/10.1186/s13640-020-0496-6 |
| work_keys_str_mv | AT yueyuan ascaleadaptiveobjecttrackingalgorithmwithocclusiondetection AT junchu ascaleadaptiveobjecttrackingalgorithmwithocclusiondetection AT luleng ascaleadaptiveobjecttrackingalgorithmwithocclusiondetection AT junmiao ascaleadaptiveobjecttrackingalgorithmwithocclusiondetection AT byunggyukim ascaleadaptiveobjecttrackingalgorithmwithocclusiondetection AT yueyuan scaleadaptiveobjecttrackingalgorithmwithocclusiondetection AT junchu scaleadaptiveobjecttrackingalgorithmwithocclusiondetection AT luleng scaleadaptiveobjecttrackingalgorithmwithocclusiondetection AT junmiao scaleadaptiveobjecttrackingalgorithmwithocclusiondetection AT byunggyukim scaleadaptiveobjecttrackingalgorithmwithocclusiondetection |