Journal Cover – Impact in Agriculture

Impact in Agriculture

Peer-Reviewed • Open Access e-ISSN: 3122-735X

Submit Your Manuscript
Article Open Access 800 Views | 120 Downloads | 1–16 | PDF

Impact of Dataset Quality on Deep Learning Models for Dragon Fruit and Leaf Health Classification

Shahnawaz Ayoub 1 ORCID , Imran Baig 2 ORCID , Mudasir Ashraf 3 , Mahmoud Okasha 4,5 ORCID
1 Glocal School of Science and Technology, Glocal University, Saharanpur, Uttar Pradesh 247121, India
2 Cardiff School of Technologies, Cardiff Metropolitan University, Llandaff Campus, Western Avenue, Cardiff CF5 2YB, The UK
3 School of engineering and IT, Manipal academy of Higher Education, Dubai, 345050, United Arab Emirates
4 Agricultural Engineering Research Institute, Agricultural Research Center, Dokki, Giza 12611, Egypt
5 Department of Agricultural, Food, Environmental and Animal Sciences, University of Udine, Udine I-33100, Italy
DOI: https://doi.org/10.65500/agriculture-2025-001
Received: 16 August 2025 | Revised: 17 September 2025 | Accepted: 25 September 2025 | Published: 13 October 2025

Abstract

Accurate assessment of fruit and leaf health is essential for early disease detection, quality grading, and automated management in commercial dragon fruit production. Variability in illumination, symptom intensity, and morphological features often limits the reliability of conventional machine learning models trained on raw datasets. This study evaluates the effect of dataset quality on deep learning performance using a publicly available dragon fruit and leaf dataset containing 4,518 images across four classes: Healthy Fruit, Healthy Leaves, Infected Fruits, and Infected Leaves. Three dataset versions were constructed (i) the original dataset, (ii) an augmented dataset expanding each image threefold, and (iii) a cleaned augmented dataset created by removing mislabeled, ambiguous, or low-quality samples. Four deep architectures (MobileNetV3, InceptionV3, ResNet101, and VGG16) were trained under identical settings to assess classification performance. Across all models, the cleaned augmented dataset produced the most stable training behavior and highest accuracy. InceptionV3 achieved the strongest overall performance with an F1-score above 0.95 and validation accuracy approaching 0.97, while MobileNetV3 delivered competitive results (accuracy 0.9613) with minimal computational cost. Confusion matrices confirmed major reductions in fruit–fruit and leaf–leaf misclassification after dataset cleaning. The findings highlight that targeted data refinement, combined with augmentation, is critical for building reliable deep learning models for real-world agricultural applications.

Keywords: dragon fruit classification; deep learning; dataset augmentation; image cleaning; MobileNetV3; InceptionV3; plant disease detection

References

  1. Luu, T.T.H.; Le, T.L.; Huynh, N.; Quintela-Alonso, P. Dragon Fruit: A Review of Health Benefits and Nutrients and Its Sustainable Development under Climate Changes in Vietnam. https://cjfs.agriculturejournals.cz/doi/10.17221/139/2020-CJFS.html 2021, 39, 71–94, doi:10.17221/139/2020-CJFS.
  2. Trivellini, A.; Lucchesini, M.; Ferrante, A.; Massa, D.; Orlando, M.; Incrocci, L.; Mensuali-Sodi, A. Pitaya, an Attractive Alternative Crop for Mediterranean Region. Agronomy 2020, Vol. 10, Page 1065 2020, 10, 1065, doi:10.3390/AGRONOMY10081065.
  3. Raju, C.; Pazhanivelan, S.; Perianadar, I.V.; Kaliaperumal, R.; Sathyamoorthy, N.K.; Sendhilvel, V. Climate Change as an Existential Threat to Tropical Fruit Crop Production—A Review. Agriculture 2024, Vol. 14, Page 2018 2024, 14, 2018, doi:10.3390/AGRICULTURE14112018.
  4. Amri, E.; Gulzar, Y.; Yeafi, A.; Jendoubi, S.; Dhawi, F.; Mir, M.S. Advancing Automatic Plant Classification System in Saudi Arabia: Introducing a Novel Dataset and Ensemble Deep Learning Approach. Model Earth Syst Environ 2024, 10, 2693–2709, doi:10.1007/S40808-023-01918-9/METRICS.
  5. Gulzar, Y.; Ünal, Z. Optimizing Pear Leaf Disease Detection Through PL-DenseNet. Applied Fruit Science 2025, 67, 1–13, doi:10.1007/s10341-025-01265-2.
  6. Gulzar, Y.; Ünal, Z. Time-Sensitive Bruise Detection in Plums Using PlmNet with Transfer Learning. Procedia Comput Sci 2025, 257, 127–132, doi:10.1016/J.PROCS.2025.03.019.
  7. Gulzar, Y. Fruit Image Classification Model Based on MobileNetV2 with Deep Transfer Learning Technique. Sustainability 2023, 15, 1906.
  8. Kulkarni, V.; Kosamkar, P.; Singh, C.; Ingle, P.; Modi, V. Detection and Classification of Diseases and Maturity of Dragon Fruits. Lecture Notes in Networks and Systems 2022, 321, 365–374, doi:10.1007/978-981-16-5987-4_37.
  9. Saranya, T.; Deisy, C.; Sridevi, S. Efficient Agricultural Pest Classification Using Vision Transformer with Hybrid Pooled Multihead Attention. Comput Biol Med 2024, 177, 108584, doi:10.1016/J.COMPBIOMED.2024.108584.
  10. Gulzar, Y. Papaya Leaf Disease Classification Using Pre-Trained Deep Learning Models: A Comparative Study. Applied Fruit Science 2025, 67, 1–10, doi:10.1007/S10341-025-01533-1/METRICS.
  11. Gulzar, Y. PapNet: An AI-Driven Approach for Early Detection and Classification of Papaya Leaf Diseases. Applied Fruit Science 2025 67:4 2025, 67, 1–11, doi:10.1007/S10341-025-01466-9.
  12. Sarkar, P.; Pranta, G.K.; Mojumdar, M.U. Dragon Fruit & Leaf Dataset from Bangladesh for Classification and Ecological Research. 2024, 1, doi:10.17632/CFCHFDPFW5.1.
  13. Khan, A.; Radzi, S.A.; Zaimi, M.Z.M.; Amsan, A.N.; Mohd Saad, W.H.; Abd Razak, N.A.; Hamid, N.A.; Samad, A.S.A. Revolutionizing Agriculture with Deep Learning Current Trends and Future Directions. International Journal of Integrated Engineering 2024, 16, 192–211, doi:10.30880/ijie.2024.16.03.018.
  14. Zhang, W.; Zheng, C.; Wang, C.; Guo, W. DomAda-FruitDet: Domain-Adaptive Anchor-Free Fruit Detection Model for Auto Labeling. Plant Phenomics 2024, 6, doi:10.34133/plantphenomics.0135.
  15. Zhang, W.; Chen, K.; Zheng, C.; Liu, Y.; Guo, W. EasyDAM_V2: Efficient Data Labeling Method for Multishape, Cross-Species Fruit Detection. Plant Phenomics 2022, 2022, doi:10.34133/2022/9761674.
  16. Rahmania, R.; Corputty, F.; Wibowo, S.A.; Saputra, D.E.; Arrahmah, A.I. Exploration of The Impact of Kernel Size for YOLOv5-Based Object Detection on Quadcopter. International Journal on Informatics Visualization 2022, 6, 726–735, doi:10.30630/joiv.6.3.898.
  17. Minh Trieu, N.; Thinh, N.T. Quality Classification of Dragon Fruits Based on External Performance Using a Convolutional Neural Network. Applied Sciences (Switzerland) 2021, 11, doi:10.3390/app112210558.
  18. Patil, P.U.; Lande, S.B.; Nagalkar, V.J.; Nikam, S.B.; Wakchaure, G.C. Grading and Sorting Technique of Dragon Fruits Using Machine Learning Algorithms. J Agric Food Res 2021, 4, doi:10.1016/j.jafr.2021.100118.
  19. Yusamran, N.; Hiransakolwong, N. DIPDEEP: Classification for Thai Dragon Fruit. Engineering and Applied Science Research 2022, 49, 521–530.
  20. Pallavi, N.; Vijayakarthik, P.; Sushma, B. Optimizing Dragon Fruit Quality and Maturity Classification Through Deep Learning Techniques. SN Comput Sci 2025, 6, doi:10.1007/s42979-025-04195-8.
  21. Abhishek, A.G.S.; Ravikumar, T.; Terlapu, P.V.; Tippana, C.; Pondreti, R. Intelligent Fruit Detection System Using Optimized Hybrid Deep Learning Models. Journal of Machine and Computing 2025, 5, 1386–1395, doi:10.53759/7669/jmc202505109.
  22. Nikam, S.B.; Lande, S.B.; Nagalkar, V.J.; Wakchaure, G.C.; Kumar, P.S. Predictive Classification Model for Quality Grading and Maturity Detection of Dragon Fruit Using Fused Deep CNN Feature and Ensemble Learning. J Food Process Preserv 2025, 2025, doi:10.1155/jfpp/6938071.
  23. da Silva-Ferreira, M.V.; Barbon Junior, S.; Turrisi da Costa, V.G.; Barbin, D.F.; Lucena-Barbosa, J.E. De Deep Computer Vision System and Explainable Artificial Intelligence Applied for Classification of Dragon Fruit (Hylocereus Spp.). Sci Hortic 2024, 338, doi:10.1016/j.scienta.2024.113605.
  24. Vo, H.T.; Thien, N.N.; Mui, K.C. A Deep Transfer Learning Approach for Accurate Dragon Fruit Ripeness Classification and Visual Explanation Using Grad-CAM. International Journal of Advanced Computer Science and Applications 2023, 14, 1344–1352, doi:10.14569/IJACSA.2023.01411137.
  25. Cometa, L.M.A.; Garcia, R.K.T.; Latina, M.A.E. Real-Time Visual Identification System to Assess Maturity, Size, and Defects in Dragon Fruits †. Engineering Proceedings 2025, 92, doi:10.3390/engproc2025092039.
  26. Khatun, T.; Nirob, M.A.S.; Bishshash, P.; Akter, M.; Uddin, M.S. A Comprehensive Dragon Fruit Image Dataset for Detecting the Maturity and Quality Grading of Dragon Fruit. Data Brief 2024, 52, doi:10.1016/j.dib.2023.109936.
  27. Li, X.; Wang, X.; Ong, P.; Yi, Z.; Ding, L.; Han, C. Fast Recognition and Counting Method of Dragon Fruit Flowers and Fruits Based on Video Stream. Sensors 2023, 23, doi:10.3390/s23208444.
  28. Ha, D.M.; Hung, T.; Kieu, N.X.; Vuong, N.G.; Thuy, Q.D.T. Semantic Connection-Based Learning for Dragon Fruit Disease Classification. Journal of Information Hiding and Multimedia Signal Processing 2024, 15, 281–291.
  29. Nguyen, T.P.T.; Nguyen, T.T.; Nguyen, H.Q.; Nguyen, T.D.; Nguyen, C.K.; Cu, N.G. An Enhanced Image Classification Model Based on Graph Classification and Superpixel-Derived CNN Features for Agricultural Datasets. Computers, Materials and Continua 2025, 85, 4899–4920, doi:10.32604/cmc.2025.067707.
  30. Wang, J.; Gao, K.; Jiang, H.; Zhou, H. Method for Detecting Dragon Fruit Based on Improved Lightweight Convolutional Neural Network; 基于改进的轻量化卷积神经网络火龙果检测方法. Nongye Gongcheng Xuebao/Transactions of the Chinese Society of Agricultural Engineering 2020, 36, 218–225, doi:10.11975/j.issn.1002-6819.2020.20.026.
  31. Shang, F.; Zhou, X.; Liang, Y.; Xiao, M.; Chen, Q.; Luo, C. Detection Method for Dragon Fruit in Natural Environment Based on Improved YOLOX; 基于改进 YOLOX 的自然环境中火龙果检测方法. Smart Agriculture 2022, 4, 120–131, doi:10.12133/j.smartag.SA202207001.
  32. Wang, J.; Zhou, J.; Zhang, Y.; Hu, H. Multi-Pose Dragon Fruit Detection System for Picking Robots Based on the Optimal YOLOv7 Model; 基于优选 YOLOv7 模型的采摘机器人多姿态火龙果检测系统. Nongye Gongcheng Xuebao/Transactions of the Chinese Society of Agricultural Engineering 2023, 39, 276–283, doi:10.11975/j.issn.1002-6819.202208031.
  33. Zhou, J.; Zhang, Y.; Wang, J. A Dragon Fruit Picking Detection Method Based on YOLOv7 and PSP-Ellipse. Sensors 2023, 23, doi:10.3390/s23083803.
  34. Zhu, L.; Deng, W.; Lai, Y.; Guo, X.; Zhang, S. Research on Improved Road Visual Navigation Recognition Method Based on DeepLabV3+ in Pitaya Orchard. Agronomy 2024, 14, doi:10.3390/agronomy14061119.
  35. Zhou, Z.; Peng, R.; Li, R.; Li, Y.; Huang, D.; Zhu, M. Remote Sensing Identification and Rapid Yield Estimation of Pitaya Plants in Different Karst Mountainous Complex Habitats. Agriculture (Switzerland) 2023, 13, doi:10.3390/agriculture13091742.
  36. Li, Q.; Yan, L.; Huang, D.; Zhou, Z.; Zhang, Y.; Xiao, D. Construction of a Small Sample Dataset and Identification of Pitaya Trees (Selenicereus) Based on UAV Image on Close-Range Acquisition. J Appl Remote Sens 2022, 16, doi:10.1117/1.JRS.16.024502.
  37. Yu, J.; Sun, Y.; Latinovic, N.; Kong, C.; Han, B.; Zhang, X. Nondestructive Internal Quality Detection Method for Yellow Pitaya Based on EIS and Tactile Multimodal Perception Data-Driven Approach. Journal of Food Composition and Analysis 2025, 144, doi:10.1016/j.jfca.2025.107744.
  38. Pan, Y.; Wang, Y.; Zhou, Y.; Zhou, J.; Chen, M.; Liu, D.; Li, F.; Liu, C.; Zeng, M.; Jiang, D.; et al. A Smartphone-Based Non-Destructive Multimodal Deep Learning Approach Using PH-Sensitive Pitaya Peel Films for Real-Time Fish Freshness Detection. Foods 2025, 14, doi:10.3390/foods14101805.
  39. Xu, T.; Song, L.; Lu, X.; Zhang, H. Dual-Index Detection Method of Pitaya Quality and Maturity Based on YOLO v7-RA; 基于YOLO v7 RA 的火龙果品质与成熟度双指标检测方法. Nongye Jixie Xuebao/Transactions of the Chinese Society for Agricultural Machinery 2024, 55, 405–414, doi:10.6041/j.issn.1000-1298.2024.07.040.
© 2026 by the authors. Published by Impaxon Publishing.
This article is licensed under the Creative Commons Attribution (CC BY) License .
You are free to share and adapt the material as long as appropriate credit is given.
Publisher’s Note: All claims expressed in this article are solely those of the authors and do not necessarily represent those of Impaxon Publishing or the journal editors. The publisher remains neutral with regard to jurisdictional claims in institutional affiliations.