Semi-Supervised Semantic Segmentation for Identification of Irrelevant Objects in a Waste Recycling Plant
In waste recycling plants, measuring the waste volume and weight at the beginning of the treatment process is key for a better management of resources. This task can be conducted by using orthophoto images, but it is necessary to remove from those images the objects, such as containers or trucks, th...
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Graz University of Technology
2023-05-01
|
| Series: | Journal of Universal Computer Science |
| Subjects: | |
| Online Access: | https://lib.jucs.org/article/87643/download/pdf/ |
| _version_ | 1797816791640047616 |
|---|---|
| author | César Domínguez Jónathan Heras Eloy Mata Vico Pascual Lucas Fernández-Cedrón Marcos Martínez-Lanchares Jon Pellejero-Espinosa Antonio Rubio-Loscertales Carlos Tarragona-Pérez |
| author_facet | César Domínguez Jónathan Heras Eloy Mata Vico Pascual Lucas Fernández-Cedrón Marcos Martínez-Lanchares Jon Pellejero-Espinosa Antonio Rubio-Loscertales Carlos Tarragona-Pérez |
| author_sort | César Domínguez |
| collection | DOAJ |
| description | In waste recycling plants, measuring the waste volume and weight at the beginning of the treatment process is key for a better management of resources. This task can be conducted by using orthophoto images, but it is necessary to remove from those images the objects, such as containers or trucks, that are not involved in the measurement process. This work proposes the application of deep learning for the semantic segmentation of those irrelevant objects. Several deep architectures are trained and compared, while three semi-supervised learning methods (PseudoLabeling, Distillation and Model Distillation) are proposed to take advantage of non-annotated images. In these experiments, the U-net++ architecture with an EfficientNetB3 backbone, trained with the set of labelled images, achieves the best overall multi Dice score of 91.23%. The application of semi-supervised learning methods further boosts the segmentation accuracy in a range between 1.31% and 2.59%, on average. |
| first_indexed | 2024-03-13T08:42:58Z |
| format | Article |
| id | doaj.art-44b187d4b7e247da9dc48eb3f45871c7 |
| institution | Directory Open Access Journal |
| issn | 0948-6968 |
| language | English |
| last_indexed | 2024-03-13T08:42:58Z |
| publishDate | 2023-05-01 |
| publisher | Graz University of Technology |
| record_format | Article |
| series | Journal of Universal Computer Science |
| spelling | doaj.art-44b187d4b7e247da9dc48eb3f45871c72023-05-30T08:11:04ZengGraz University of TechnologyJournal of Universal Computer Science0948-69682023-05-0129541943110.3897/jucs.8764387643Semi-Supervised Semantic Segmentation for Identification of Irrelevant Objects in a Waste Recycling PlantCésar Domínguez0Jónathan Heras1Eloy Mata2Vico Pascual3Lucas Fernández-Cedrón4Marcos Martínez-Lanchares5Jon Pellejero-Espinosa6Antonio Rubio-Loscertales7Carlos Tarragona-Pérez8Universidad de La RiojaUniversidad de La RiojaUniversidad de La RiojaUniversidad de La RiojaSpectralGeoSpectralGeoSpectralGeoSpectralGeoSpectralGeoIn waste recycling plants, measuring the waste volume and weight at the beginning of the treatment process is key for a better management of resources. This task can be conducted by using orthophoto images, but it is necessary to remove from those images the objects, such as containers or trucks, that are not involved in the measurement process. This work proposes the application of deep learning for the semantic segmentation of those irrelevant objects. Several deep architectures are trained and compared, while three semi-supervised learning methods (PseudoLabeling, Distillation and Model Distillation) are proposed to take advantage of non-annotated images. In these experiments, the U-net++ architecture with an EfficientNetB3 backbone, trained with the set of labelled images, achieves the best overall multi Dice score of 91.23%. The application of semi-supervised learning methods further boosts the segmentation accuracy in a range between 1.31% and 2.59%, on average.https://lib.jucs.org/article/87643/download/pdf/Waste managementSemantic SegmentationDeep Lear |
| spellingShingle | César Domínguez Jónathan Heras Eloy Mata Vico Pascual Lucas Fernández-Cedrón Marcos Martínez-Lanchares Jon Pellejero-Espinosa Antonio Rubio-Loscertales Carlos Tarragona-Pérez Semi-Supervised Semantic Segmentation for Identification of Irrelevant Objects in a Waste Recycling Plant Journal of Universal Computer Science Waste management Semantic Segmentation Deep Lear |
| title | Semi-Supervised Semantic Segmentation for Identification of Irrelevant Objects in a Waste Recycling Plant |
| title_full | Semi-Supervised Semantic Segmentation for Identification of Irrelevant Objects in a Waste Recycling Plant |
| title_fullStr | Semi-Supervised Semantic Segmentation for Identification of Irrelevant Objects in a Waste Recycling Plant |
| title_full_unstemmed | Semi-Supervised Semantic Segmentation for Identification of Irrelevant Objects in a Waste Recycling Plant |
| title_short | Semi-Supervised Semantic Segmentation for Identification of Irrelevant Objects in a Waste Recycling Plant |
| title_sort | semi supervised semantic segmentation for identification of irrelevant objects in a waste recycling plant |
| topic | Waste management Semantic Segmentation Deep Lear |
| url | https://lib.jucs.org/article/87643/download/pdf/ |
| work_keys_str_mv | AT cesardominguez semisupervisedsemanticsegmentationforidentificationofirrelevantobjectsinawasterecyclingplant AT jonathanheras semisupervisedsemanticsegmentationforidentificationofirrelevantobjectsinawasterecyclingplant AT eloymata semisupervisedsemanticsegmentationforidentificationofirrelevantobjectsinawasterecyclingplant AT vicopascual semisupervisedsemanticsegmentationforidentificationofirrelevantobjectsinawasterecyclingplant AT lucasfernandezcedron semisupervisedsemanticsegmentationforidentificationofirrelevantobjectsinawasterecyclingplant AT marcosmartinezlanchares semisupervisedsemanticsegmentationforidentificationofirrelevantobjectsinawasterecyclingplant AT jonpellejeroespinosa semisupervisedsemanticsegmentationforidentificationofirrelevantobjectsinawasterecyclingplant AT antoniorubioloscertales semisupervisedsemanticsegmentationforidentificationofirrelevantobjectsinawasterecyclingplant AT carlostarragonaperez semisupervisedsemanticsegmentationforidentificationofirrelevantobjectsinawasterecyclingplant |