SavedAdd your journal

Journals / Turkish Journal of Science and Technology / 2023 / Cilt: 18 Sayı: 2

A Hybrid Classification Approach for Fasteners Based on Transfer Learning with Fine-Tuning and Deep Features

Derin Öznitelik ve İnce-Ayar ile Aktarım Öğrenme Tabanlı Bağlantı Elemanlarının Hibrit Sınıflandırma Yaklaşımı

Canan TAŞTİMUR
Erhan AKIN
Pages
461–475
Publication Date
2023-09-01
DOI
—
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

The models with the best results are DenseNet169 with an accuracy of 0.97 in the TL-FT1 scenario, EfficientNetB0 with 0.96 in TL-FT2, and DenseNet169 with 0.995 in TL-FE.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:Deep learning, which has seen frequent use in recent studies, has helped solve the problem of classifying objects of many different types and properties.
📊
Key Findings & Data:The study compares the classification performance of eighteen different pre-trained network models (i.e., one or more versions of EfficientNet, DenseNet, InceptionResNetV2, InceptionV3, MobileNet, ResNet50, Xception, and VGGNet) in detail.
💡
Conclusion & Impact:The models with the best results are DenseNet169 with an accuracy of 0.97 in the TL-FT1 scenario, EfficientNetB0 with 0.96 in TL-FT2, and DenseNet169 with 0.995 in TL-FE.
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:Sınıflandırma · bağlantı elemanı · özellik çıkarma

Abstract

Deep learning, which has seen frequent use in recent studies, has helped solve the problem of classifying objects of many different types and properties. Most studies both create and train a convolutional neural network (CNN) from scratch. The time spent training the network is thus wasted. Transfer learning (TL) is used both to prevent the loss of time due to training the dataset and to more effectively classify small datasets. This study performs classification using a dataset containing eighteen types of fastener. Our study contains three different TL scenarios. Two of them use TL with fine-tuning (FT), while the third does so with feature extraction (FE). The study compares the classification performance of eighteen different pre-trained network models (i.e., one or more versions of EfficientNet, DenseNet, InceptionResNetV2, InceptionV3, MobileNet, ResNet50, Xception, and VGGNet) in detail. When compared to other research in the literature, our first and second scenarios provide excellent implementations of TL-FT, while our third scenario, TL-FE, is hybrid and produces better results than the other two. Furthermore, our findings are superior to those of most previous studies. The models with the best results are DenseNet169 with an accuracy of 0.97 in the TL-FT1 scenario, EfficientNetB0 with 0.96 in TL-FT2, and DenseNet169 with 0.995 in TL-FE.

Özet

Son yıllarda yapılan çalışmalarda sıkça kullanılmaya başlanan derin öğrenme, birçok farklı tür ve özellikteki nesnelerin sınıflandırılması sorununun çözülmesine yardımcı olmuştur. Çoğu çalışma, sıfırdan bir evrişimsel sinir ağı (CNN) oluşturur ve eğitir. Ağı eğitmek için harcanan zaman böylece boşa harcanır. Transfer öğrenme (TL) hem veri setinin eğitilmesinden kaynaklanan zaman kaybını önlemek hem de küçük veri setlerini daha etkin bir şekilde sınıflandırmak için kullanılmaktadır. Bu çalışma, on sekiz tip bağlantı elemanı içeren bir veri seti kullanarak sınıflandırma yapmaktadır. Çalışmamız üç farklı TL senaryosu içermektedir. Bunlardan ikisi ince ayar (FT) ile TL kullanırken, üçüncüsü özellik çıkarma (FE) ile yapmaktadır. Çalışma, on sekiz farklı önceden eğitilmiş ağ modelinin (yani EfficientNet, DenseNet, InceptionResNetV2, InceptionV3, MobileNet, ResNet50, Xception ve VGGNet) sınıflandırma performansını ayrıntılı olarak karşılaştırmaktadır. Literatürdeki diğer araştırmalarla karşılaştırıldığında, birinci ve ikinci senaryolarımız TL-FT' nin iyi sonuçlarla uygulamalarını sağlarken, üçüncü senaryomuz TL-FE hibrit bir yöntem olup diğer iki senaryodan daha iyi sonuçlar üretmiştir. Ayrıca, bulgularımız literatürdeki çalışmaların çoğundan daha üstün olduğu fark edilmiştir. En iyi sonuçlara sahip modeller TL-FT1 senaryosunda 0,97 doğrulukla DenseNet169, TL-FT2'de 0,96 ile EfficientNetB0 ve TL-FE'de 0,995 ile DenseNet169'dur.

🔖

Cite This Article

TAŞTİMUR, C., AKIN, E. (2023). A Hybrid Classification Approach for Fasteners Based on Transfer Learning with Fine-Tuning and Deep Features. Turkish Journal of Science and Technology, Cilt: 18 Sayı: 2, 461–475. https://ajindex.com/articles/959784

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (9)

  1. Pathak Y, Shukla PK, Tiwari A, Stalin S, & Singh S. Deep transfer learning based classification model for COVID-19 disease. Pattern Recognit. Lett. 2020; 152: 122-128.
  2. Akgun D, Kabakuş AT, Senturk Z.K, Senturk A, & Kucukkulahli E. A transfer learning-based deep learning approach for automated COVID-19 diagnosis with audio data. Turk. J. Electr. Eng. Comput. Sci. 2021; 29(SI-1): 2807-2823.
  3. Sravan V, Swaraj K, Meenakshi K, & Kora P. A deep learning based crop disease classification using transfer learning. Mater. Today Proc. 2020.
  4. Kudva V, Prasad K, & Guruvare S. Hybrid transfer learning for classification of uterine cervix images for cervical cancer screening. J. Digital Imaging 2020; 33(3): 619-631. Thenmozhi K, & Reddy U.S. Crop pest classification based on deep convolutional neural network and transfer learning. Comput. Electron. Agric. 2019; 164. Talo M, Baloglu UB, Yıldırım O, & Acharya UR. Application of deep transfer learning for automated brain abnormality classification using MR images. Cogn. Syst. Res. 2019; 154: 176-188.
  5. Mehrotra R, Ansari M, Agrawal R, & Anand RS. A transfer learning approach for AI-based classification of brain tumors. Mach. Learn. Appl. 2020; 2. Yang K, Yang T, Yao Y, & Fan SD. A transfer learning-based convolutional neural network and its novel application in ship spare-parts classification. Ocean Coastal Manage. 2021; 215. Ali MS, Miah M.S, Haque J, Rahman MM, & Islam MK. An enhanced technique of skin cancer classification using deep convolutional neural network with transfer learning models. Mach. Learn. Appl. 2021; 5. Rahman Z, & Ami AM. A transfer learning based approach for skin lesion classification from imbalanced data. In: 2020 11th International Conference on Electrical and Computer Engineering (ICECE), 2020;
  6. Dhaka, Bangladesh. pp. 65-68. Kumar S, & Janet B. DTMIC: Deep transfer learning for malware image classification. J. Inf. Secur. Appl. 2022; 64. Giraddi S, Seeri S, Hiremath P.S, & Jayalaxmi GN. Flower Classification using Deep Learning models. In: 2020 International Conference on Smart Technologies in Computing, Electrical and Electronics (ICSTCEE), 2020;
  7. Karnataka, India. pp. 130-133. Wang I. H. Lee KC, & Chang SL. Images Classification of Dogs and Cats using Fine-Tuned VGG Models. In: 2020
  8. IEEE Eurasia Conference on IOT, Communication and Engineering (ECICE), 2020;
  9. Yunlin, Taiwan: IEEE. pp. 230-233. Lee SW. Novel classification method of plastic wastes with optimal hyper-parameter tuning of Inception_ResnetV2