Transfer Learning Methods as a New Approach in Computer Vision Tasks with Small Datasets
Deep learning methods, used in machine vision challenges, often face the problem of the amount and quality of data. To address this issue, we investigate the transfer learning method. In this study, we briefly describe the idea and introduce two main strategies of transfer learning. We also present...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Sciendo
2020-09-01
|
| Series: | Foundations of Computing and Decision Sciences |
| Subjects: | |
| Online Access: | https://doi.org/10.2478/fcds-2020-0010 |
| id |
doaj-8376a86fc23c44d0a53ff8f6b3d6d456 |
|---|---|
| record_format |
Article |
| spelling |
doaj-8376a86fc23c44d0a53ff8f6b3d6d4562021-09-05T21:00:54ZengSciendoFoundations of Computing and Decision Sciences2300-34052020-09-0145317919310.2478/fcds-2020-0010fcds-2020-0010Transfer Learning Methods as a New Approach in Computer Vision Tasks with Small DatasetsBrodzicki Andrzej0Piekarski Michal1Kucharski Dariusz2Jaworek-Korjakowska Joanna3Gorgon Marek4Department of Automatic Control and Robotics, AGH University of Science and Technology, Krakow, PolandDepartment of Automatic Control and Robotics, AGH University of Science and Technology, Krakow, PolandDepartment of Automatic Control and Robotics, AGH University of Science and Technology, Krakow, PolandDepartment of Automatic Control and Robotics, AGH University of Science and Technology, Krakow, PolandDepartment of Automatic Control and Robotics, AGH University of Science and Technology, Krakow, PolandDeep learning methods, used in machine vision challenges, often face the problem of the amount and quality of data. To address this issue, we investigate the transfer learning method. In this study, we briefly describe the idea and introduce two main strategies of transfer learning. We also present the widely-used neural network models, that in recent years performed best in ImageNet classification challenges. Furthermore, we shortly describe three different experiments from computer vision field, that confirm the developed algorithms ability to classify images with overall accuracy 87.2-95%. Achieved numbers are state-of-the-art results in melanoma thickness prediction, anomaly detection and Clostridium di cile cytotoxicity classification problems.https://doi.org/10.2478/fcds-2020-0010deep neural networkstransfer learningsignal processingimage analysisanomaly detection |
| collection |
DOAJ |
| language |
English |
| format |
Article |
| sources |
DOAJ |
| author |
Brodzicki Andrzej Piekarski Michal Kucharski Dariusz Jaworek-Korjakowska Joanna Gorgon Marek |
| spellingShingle |
Brodzicki Andrzej Piekarski Michal Kucharski Dariusz Jaworek-Korjakowska Joanna Gorgon Marek Transfer Learning Methods as a New Approach in Computer Vision Tasks with Small Datasets Foundations of Computing and Decision Sciences deep neural networks transfer learning signal processing image analysis anomaly detection |
| author_facet |
Brodzicki Andrzej Piekarski Michal Kucharski Dariusz Jaworek-Korjakowska Joanna Gorgon Marek |
| author_sort |
Brodzicki Andrzej |
| title |
Transfer Learning Methods as a New Approach in Computer Vision Tasks with Small Datasets |
| title_short |
Transfer Learning Methods as a New Approach in Computer Vision Tasks with Small Datasets |
| title_full |
Transfer Learning Methods as a New Approach in Computer Vision Tasks with Small Datasets |
| title_fullStr |
Transfer Learning Methods as a New Approach in Computer Vision Tasks with Small Datasets |
| title_full_unstemmed |
Transfer Learning Methods as a New Approach in Computer Vision Tasks with Small Datasets |
| title_sort |
transfer learning methods as a new approach in computer vision tasks with small datasets |
| publisher |
Sciendo |
| series |
Foundations of Computing and Decision Sciences |
| issn |
2300-3405 |
| publishDate |
2020-09-01 |
| description |
Deep learning methods, used in machine vision challenges, often face the problem of the amount and quality of data. To address this issue, we investigate the transfer learning method. In this study, we briefly describe the idea and introduce two main strategies of transfer learning. We also present the widely-used neural network models, that in recent years performed best in ImageNet classification challenges. Furthermore, we shortly describe three different experiments from computer vision field, that confirm the developed algorithms ability to classify images with overall accuracy 87.2-95%. Achieved numbers are state-of-the-art results in melanoma thickness prediction, anomaly detection and Clostridium di cile cytotoxicity classification problems. |
| topic |
deep neural networks transfer learning signal processing image analysis anomaly detection |
| url |
https://doi.org/10.2478/fcds-2020-0010 |
| work_keys_str_mv |
AT brodzickiandrzej transferlearningmethodsasanewapproachincomputervisiontaskswithsmalldatasets AT piekarskimichal transferlearningmethodsasanewapproachincomputervisiontaskswithsmalldatasets AT kucharskidariusz transferlearningmethodsasanewapproachincomputervisiontaskswithsmalldatasets AT jaworekkorjakowskajoanna transferlearningmethodsasanewapproachincomputervisiontaskswithsmalldatasets AT gorgonmarek transferlearningmethodsasanewapproachincomputervisiontaskswithsmalldatasets |
| _version_ |
1717782065188962304 |