Overcoming Domain Shift in Neural Networks for Accurate Plant Counting in Aerial Images
This paper presents a novel semi-supervised approach for accurate counting and localization of tropical plants in aerial images that can work in new visual domains in which the available data are not labeled. Our approach uses deep learning and domain adaptation, designed to handle domain shifts bet...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-03-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/15/6/1700 |
| _version_ | 1797609146688733184 |
|---|---|
| author | Javier Rodriguez-Vazquez Miguel Fernandez-Cortizas David Perez-Saura Martin Molina Pascual Campoy |
| author_facet | Javier Rodriguez-Vazquez Miguel Fernandez-Cortizas David Perez-Saura Martin Molina Pascual Campoy |
| author_sort | Javier Rodriguez-Vazquez |
| collection | DOAJ |
| description | This paper presents a novel semi-supervised approach for accurate counting and localization of tropical plants in aerial images that can work in new visual domains in which the available data are not labeled. Our approach uses deep learning and domain adaptation, designed to handle domain shifts between the training and test data, which is a common challenge in this agricultural applications. This method uses a source dataset with annotated plants and a target dataset without annotations and adapts a model trained on the source dataset to the target dataset using unsupervised domain alignment and pseudolabeling. The experimental results show the effectiveness of this approach for plant counting in aerial images of pineapples under significative domain shift, achieving a reduction up to 97% in the counting error (1.42 in absolute count) when compared to the supervised baseline (48.6 in absolute count). |
| first_indexed | 2024-03-11T05:57:30Z |
| format | Article |
| id | doaj.art-a40d76507fee48eda7f8d241d9d90146 |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-11T05:57:30Z |
| publishDate | 2023-03-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-a40d76507fee48eda7f8d241d9d901462023-11-17T13:40:46ZengMDPI AGRemote Sensing2072-42922023-03-01156170010.3390/rs15061700Overcoming Domain Shift in Neural Networks for Accurate Plant Counting in Aerial ImagesJavier Rodriguez-Vazquez0Miguel Fernandez-Cortizas1David Perez-Saura2Martin Molina3Pascual Campoy4Computer Vision and Aerial Robotics Group, Centre for Automation and Robotics (C.A.R.), Universidad Politécnica de Madrid (UPM-CSIC), 28006 Madrid, SpainComputer Vision and Aerial Robotics Group, Centre for Automation and Robotics (C.A.R.), Universidad Politécnica de Madrid (UPM-CSIC), 28006 Madrid, SpainComputer Vision and Aerial Robotics Group, Centre for Automation and Robotics (C.A.R.), Universidad Politécnica de Madrid (UPM-CSIC), 28006 Madrid, SpainComputer Vision and Aerial Robotics Group, Department of Artificial Intelligence, Universidad Politécnica de Madrid (UPM), 28031 Madrid, SpainComputer Vision and Aerial Robotics Group, Centre for Automation and Robotics (C.A.R.), Universidad Politécnica de Madrid (UPM-CSIC), 28006 Madrid, SpainThis paper presents a novel semi-supervised approach for accurate counting and localization of tropical plants in aerial images that can work in new visual domains in which the available data are not labeled. Our approach uses deep learning and domain adaptation, designed to handle domain shifts between the training and test data, which is a common challenge in this agricultural applications. This method uses a source dataset with annotated plants and a target dataset without annotations and adapts a model trained on the source dataset to the target dataset using unsupervised domain alignment and pseudolabeling. The experimental results show the effectiveness of this approach for plant counting in aerial images of pineapples under significative domain shift, achieving a reduction up to 97% in the counting error (1.42 in absolute count) when compared to the supervised baseline (48.6 in absolute count).https://www.mdpi.com/2072-4292/15/6/1700deep learningaerial imageryprecision agricultureplant detectiondomain adaptationunsupervised learning |
| spellingShingle | Javier Rodriguez-Vazquez Miguel Fernandez-Cortizas David Perez-Saura Martin Molina Pascual Campoy Overcoming Domain Shift in Neural Networks for Accurate Plant Counting in Aerial Images Remote Sensing deep learning aerial imagery precision agriculture plant detection domain adaptation unsupervised learning |
| title | Overcoming Domain Shift in Neural Networks for Accurate Plant Counting in Aerial Images |
| title_full | Overcoming Domain Shift in Neural Networks for Accurate Plant Counting in Aerial Images |
| title_fullStr | Overcoming Domain Shift in Neural Networks for Accurate Plant Counting in Aerial Images |
| title_full_unstemmed | Overcoming Domain Shift in Neural Networks for Accurate Plant Counting in Aerial Images |
| title_short | Overcoming Domain Shift in Neural Networks for Accurate Plant Counting in Aerial Images |
| title_sort | overcoming domain shift in neural networks for accurate plant counting in aerial images |
| topic | deep learning aerial imagery precision agriculture plant detection domain adaptation unsupervised learning |
| url | https://www.mdpi.com/2072-4292/15/6/1700 |
| work_keys_str_mv | AT javierrodriguezvazquez overcomingdomainshiftinneuralnetworksforaccurateplantcountinginaerialimages AT miguelfernandezcortizas overcomingdomainshiftinneuralnetworksforaccurateplantcountinginaerialimages AT davidperezsaura overcomingdomainshiftinneuralnetworksforaccurateplantcountinginaerialimages AT martinmolina overcomingdomainshiftinneuralnetworksforaccurateplantcountinginaerialimages AT pascualcampoy overcomingdomainshiftinneuralnetworksforaccurateplantcountinginaerialimages |