Tomato Disease Recognition Using Pseudo-Labelling-Based Semi-Supervised Learning

Strategic Insights -

Research from University of Peradeniya has led to a novel semi-supervised learning framework for automatic tomato leaf disease recognition using pseudo-labelling. The framework, named TOM-SSL, addresses the challenge of limited labelled data by leveraging a small labelled subset and confidently pseudo-labelled samples from a large pool of unlabelled data. The proposed approach achieves recognition performance on par with current state-of-the-art supervised methods while offering a tenfold enhancement in label efficiency.

Key Takeaways:

  • The availability of labelled data for disease recognition tasks is often limited due to the cost and expertise required for annotation.
  • The proposed TOM-SSL framework utilizes only 10% of the labelled data and achieves the best accuracy at 72.51% on the tomato subset of the PlantVillage dataset and 70.87% on the Taiwan tomato leaf disease dataset.
  • The framework leverages a small labelled subset and confidently pseudo-labelled samples from a large pool of unlabelled data to improve classification performance.
  • The proposed approach offers a tenfold enhancement in label efficiency compared to current state-of-the-art supervised methods.
  • The researchers used the MobileNetV3-Small backbone in the proposed framework and achieved the best performance.
  • The study highlights the potential of semi-supervised learning for automatic disease recognition in agriculture.

Statistics:

  • The proposed TOM-SSL framework achieves 72.51% accuracy on the tomato subset of the PlantVillage dataset.
  • The framework achieves 70.87% accuracy on the Taiwan tomato leaf disease dataset.
  • The proposed approach offers a tenfold enhancement in label efficiency compared to current state-of-the-art supervised methods.
  • The researchers used 10% of the labelled data in the proposed framework.
  • The study used the PlantVillage dataset, which contains 10 disease categories, and the Taiwan tomato leaf disease dataset, which contains 6 disease categories.

Sources:

  • Nishankar SN, Mithuran T, Thuseethan S, et al. TOM-SSL: Tomato Disease Recognition Using Pseudo-Labelling-Based Semi-Supervised Learning. AgriEngineering. 2025;7(8):248. Doi: 10.3390/agriengineering7080248.
  • NewsRx. Research from University of Peradeniya in the Area of Agriculture Published (TOM-SSL: Tomato Disease Recognition Using Pseudo-Labelling-Based Semi-Supervised Learning). Journal of Engineering. September 8, 2025; p 1907.