Generative Adversarial Network for Overcoming Occlusion in Images: A Survey
Although current computer vision systems are closer to the human intelligence when it comes to comprehending the visible world than previously, their performance is hindered when objects are partially occluded. Since we live in a dynamic and complex environment, we encounter more occluded objects th...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-03-01
|
| Series: | Algorithms |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1999-4893/16/3/175 |
| _version_ | 1797613961203417088 |
|---|---|
| author | Kaziwa Saleh Sándor Szénási Zoltán Vámossy |
| author_facet | Kaziwa Saleh Sándor Szénási Zoltán Vámossy |
| author_sort | Kaziwa Saleh |
| collection | DOAJ |
| description | Although current computer vision systems are closer to the human intelligence when it comes to comprehending the visible world than previously, their performance is hindered when objects are partially occluded. Since we live in a dynamic and complex environment, we encounter more occluded objects than fully visible ones. Therefore, instilling the capability of amodal perception into those vision systems is crucial. However, overcoming occlusion is difficult and comes with its own challenges. The generative adversarial network (GAN), on the other hand, is renowned for its generative power in producing data from a random noise distribution that approaches the samples that come from real data distributions. In this survey, we outline the existing works wherein GAN is utilized in addressing the challenges of overcoming occlusion, namely amodal segmentation, amodal content completion, order recovery, and acquiring training data. We provide a summary of the type of GAN, loss function, the dataset, and the results of each work. We present an overview of the implemented GAN architectures in various applications of amodal completion. We also discuss the common objective functions that are applied in training GAN for occlusion-handling tasks. Lastly, we discuss several open issues and potential future directions. |
| first_indexed | 2024-03-11T07:02:55Z |
| format | Article |
| id | doaj.art-d0c56b4a8677469bb33dad23e26287b0 |
| institution | Directory Open Access Journal |
| issn | 1999-4893 |
| language | English |
| last_indexed | 2024-03-11T07:02:55Z |
| publishDate | 2023-03-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Algorithms |
| spelling | doaj.art-d0c56b4a8677469bb33dad23e26287b02023-11-17T09:09:39ZengMDPI AGAlgorithms1999-48932023-03-0116317510.3390/a16030175Generative Adversarial Network for Overcoming Occlusion in Images: A SurveyKaziwa Saleh0Sándor Szénási1Zoltán Vámossy2Doctoral School of Applied Informatics and Applied Mathematics, Óbuda University, 1034 Budapest, HungaryJohn von Neumann Faculty of Informatics, Óbuda University, 1034 Budapest, HungaryJohn von Neumann Faculty of Informatics, Óbuda University, 1034 Budapest, HungaryAlthough current computer vision systems are closer to the human intelligence when it comes to comprehending the visible world than previously, their performance is hindered when objects are partially occluded. Since we live in a dynamic and complex environment, we encounter more occluded objects than fully visible ones. Therefore, instilling the capability of amodal perception into those vision systems is crucial. However, overcoming occlusion is difficult and comes with its own challenges. The generative adversarial network (GAN), on the other hand, is renowned for its generative power in producing data from a random noise distribution that approaches the samples that come from real data distributions. In this survey, we outline the existing works wherein GAN is utilized in addressing the challenges of overcoming occlusion, namely amodal segmentation, amodal content completion, order recovery, and acquiring training data. We provide a summary of the type of GAN, loss function, the dataset, and the results of each work. We present an overview of the implemented GAN architectures in various applications of amodal completion. We also discuss the common objective functions that are applied in training GAN for occlusion-handling tasks. Lastly, we discuss several open issues and potential future directions.https://www.mdpi.com/1999-4893/16/3/175amodal completionamodal content completionamodal segmentationamodal perceptionorder recoveryocclusion relationship |
| spellingShingle | Kaziwa Saleh Sándor Szénási Zoltán Vámossy Generative Adversarial Network for Overcoming Occlusion in Images: A Survey Algorithms amodal completion amodal content completion amodal segmentation amodal perception order recovery occlusion relationship |
| title | Generative Adversarial Network for Overcoming Occlusion in Images: A Survey |
| title_full | Generative Adversarial Network for Overcoming Occlusion in Images: A Survey |
| title_fullStr | Generative Adversarial Network for Overcoming Occlusion in Images: A Survey |
| title_full_unstemmed | Generative Adversarial Network for Overcoming Occlusion in Images: A Survey |
| title_short | Generative Adversarial Network for Overcoming Occlusion in Images: A Survey |
| title_sort | generative adversarial network for overcoming occlusion in images a survey |
| topic | amodal completion amodal content completion amodal segmentation amodal perception order recovery occlusion relationship |
| url | https://www.mdpi.com/1999-4893/16/3/175 |
| work_keys_str_mv | AT kaziwasaleh generativeadversarialnetworkforovercomingocclusioninimagesasurvey AT sandorszenasi generativeadversarialnetworkforovercomingocclusioninimagesasurvey AT zoltanvamossy generativeadversarialnetworkforovercomingocclusioninimagesasurvey |