Weakly supervised segment annotation via expectation kernel density estimation
Since the labelling for the positive images/videos is ambiguous in weakly supervised segment annotation, negative mining‐based methods that only use the intra‐class information emerge. In these methods, negative instances are utilised to penalise unknown instances for ranking their likelihood of bei...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2019-06-01
|
| Series: | IET Computer Vision |
| Subjects: | |
| Online Access: | https://doi.org/10.1049/iet-cvi.2018.5325 |
| _version_ | 1797684803182526464 |
|---|---|
| author | Liantao Wang Qingwu Li Jianfeng Lu |
| author_facet | Liantao Wang Qingwu Li Jianfeng Lu |
| author_sort | Liantao Wang |
| collection | DOAJ |
| description | Since the labelling for the positive images/videos is ambiguous in weakly supervised segment annotation, negative mining‐based methods that only use the intra‐class information emerge. In these methods, negative instances are utilised to penalise unknown instances for ranking their likelihood of being an object, which can be considered as voting in terms of similarity. However, these methods (i) ignore the information contained in positive bags; (ii) only rank the likelihood but cannot generate an explicit decision function. In this study, the authors propose a voting scheme involving not only the definite negative instances but also the ambiguous positive instances to make use of the extra useful information in the weakly labelled positive bags. In the scheme, each instance votes for its label with a magnitude arising from the similarity, and the ambiguous positive instances are assigned soft labels that are iteratively updated during the voting. It overcomes the limitations of voting using only the negative bags. They also propose an expectation kernel density estimation algorithm to gain further insight into the voting mechanism. Experimental results demonstrate the superiority of the authors’ scheme beyond the baselines. |
| first_indexed | 2024-03-12T00:35:04Z |
| format | Article |
| id | doaj.art-585de5d86fb943d1bce4b931d2258797 |
| institution | Directory Open Access Journal |
| issn | 1751-9632 1751-9640 |
| language | English |
| last_indexed | 2024-03-12T00:35:04Z |
| publishDate | 2019-06-01 |
| publisher | Wiley |
| record_format | Article |
| series | IET Computer Vision |
| spelling | doaj.art-585de5d86fb943d1bce4b931d22587972023-09-15T09:56:29ZengWileyIET Computer Vision1751-96321751-96402019-06-0113443544110.1049/iet-cvi.2018.5325Weakly supervised segment annotation via expectation kernel density estimationLiantao Wang0Qingwu Li1Jianfeng Lu2College of Internet of Things Engineering, Hohai UniversityChangzhou213022People's Republic of ChinaCollege of Internet of Things Engineering, Hohai UniversityChangzhou213022People's Republic of ChinaJiangsu Key Laboratory of Image and Video Understanding for Social SafetyNanjing University of Science and TechnologyNanjing210094People's Republic of ChinaSince the labelling for the positive images/videos is ambiguous in weakly supervised segment annotation, negative mining‐based methods that only use the intra‐class information emerge. In these methods, negative instances are utilised to penalise unknown instances for ranking their likelihood of being an object, which can be considered as voting in terms of similarity. However, these methods (i) ignore the information contained in positive bags; (ii) only rank the likelihood but cannot generate an explicit decision function. In this study, the authors propose a voting scheme involving not only the definite negative instances but also the ambiguous positive instances to make use of the extra useful information in the weakly labelled positive bags. In the scheme, each instance votes for its label with a magnitude arising from the similarity, and the ambiguous positive instances are assigned soft labels that are iteratively updated during the voting. It overcomes the limitations of voting using only the negative bags. They also propose an expectation kernel density estimation algorithm to gain further insight into the voting mechanism. Experimental results demonstrate the superiority of the authors’ scheme beyond the baselines.https://doi.org/10.1049/iet-cvi.2018.5325weakly supervised segment annotationlabellingpositive images/videosnegative mining-based methodsintra-class informationunknown instances |
| spellingShingle | Liantao Wang Qingwu Li Jianfeng Lu Weakly supervised segment annotation via expectation kernel density estimation IET Computer Vision weakly supervised segment annotation labelling positive images/videos negative mining-based methods intra-class information unknown instances |
| title | Weakly supervised segment annotation via expectation kernel density estimation |
| title_full | Weakly supervised segment annotation via expectation kernel density estimation |
| title_fullStr | Weakly supervised segment annotation via expectation kernel density estimation |
| title_full_unstemmed | Weakly supervised segment annotation via expectation kernel density estimation |
| title_short | Weakly supervised segment annotation via expectation kernel density estimation |
| title_sort | weakly supervised segment annotation via expectation kernel density estimation |
| topic | weakly supervised segment annotation labelling positive images/videos negative mining-based methods intra-class information unknown instances |
| url | https://doi.org/10.1049/iet-cvi.2018.5325 |
| work_keys_str_mv | AT liantaowang weaklysupervisedsegmentannotationviaexpectationkerneldensityestimation AT qingwuli weaklysupervisedsegmentannotationviaexpectationkerneldensityestimation AT jianfenglu weaklysupervisedsegmentannotationviaexpectationkerneldensityestimation |