Abstract
Accurate and timely identification of plant diseases along with their severity is critical for effective crop management and minimizing agricultural losses. While recent advances in deep learning have demonstrated high performance in plant disease classification, limited attention has been given to quantifying disease severity, which is essential for informed agronomic decision-making. To address this gap, this study proposes TomatoMTL, a unified multi-task learning framework for simultaneous disease classification and severity estimation of tomato leaf diseases from a single image. The proposed architecture employs a shared ResNet50-based convolutional backbone augmented with CBAM-based feature refinement, followed by task-specific branches for disease classification and severity prediction. Furthermore, a cross-task attention mechanism is introduced to enable interaction between disease-specific and severity-related features, thereby enhancing the robustness of severity estimation. To effectively leverage partially labeled data, a masking strategy is incorporated during training. Experimental evaluation on a publicly available tomato leaf disease severity dataset demonstrates that the proposed model achieves 97.85% disease classification accuracy and 77.66% severity prediction accuracy, outperforming state-of-the-art single-task classifiers including EfficientNetV2-S, ViT-B/16, and ConvNeXt-Tiny as well as existing multi-task learning baselines including Cross-Stitch Networks and MTAN. Comprehensive ablation studies confirm the individual contributions of CBAM, MixUp and CutMix augmentation, and the cross-task attention mechanism. Statistical significance analysis across five independent runs yields p-values less than 0.001 and Cohen's d greater than 14, establishing the reliability of the reported improvements. Quantitative localization analysis reveals that the model achieves 89.4% Pointing Game accuracy, confirming that attention maps focus on biologically meaningful disease regions. The proposed framework represents a complete and effective approach for integrated plant disease analysis with strong potential for real-world precision agriculture applications.
Abstract from the paper’s OpenAlex record; open copy licensed cc-by-nc-nd. Read at the source ↗
The text the model read, with the source record’s inline markup (italics, sub- and superscripts) rendered and its character entities decoded; quote offsets refer to the raw text.
Claims 9
Claim · model-written summary
TomatoMTL achieved 97.85% disease classification accuracy.
- Taxon
- Tomato
Solanum lycopersicumt0 - Outcome
- disease classification accuracye1
- Value
- 97.85%e1
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
t0“Tomato leaf disease and severity prediction”
e1“achieves 97.85% disease classification accuracy”
Claim · model-written summary
TomatoMTL outperformed single-task classifiers (EfficientNetV2-S, ViT-B/16, ConvNeXt-Tiny) in accuracy.
- Taxon
- TomatoSolanum lycopersicumt0
- Outcome
- accuracye1
- Direction
- outperforminge2
- Comparator
- state-of-the-art single-task classifiers including EfficientNetV2-S, ViT-B/16, and ConvNeXt-Tinye2
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
t0“Tomato leaf disease and severity prediction”
e1“achieves 97.85% disease classification accuracy”
e2“outperforming state-of-the-art single-task classifiers including EfficientNetV2-S, ViT-B/16, and ConvNeXt-Tiny”
Claim · model-written summary
TomatoMTL achieved 77.66% severity prediction accuracy.
- Taxon
- TomatoSolanum lycopersicumt0
- Outcome
- severity prediction accuracye1
- Value
- 77.66%e1
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
t0“Tomato leaf disease and severity prediction”
e1“77.66% severity prediction accuracy”
Claim · model-written summary
TomatoMTL outperformed existing multi-task learning baselines (Cross-Stitch Networks, MTAN) in accuracy.
- Taxon
- TomatoSolanum lycopersicumt0
- Outcome
- accuracye1
- Direction
- outperforminge2
- Comparator
- existing multi-task learning baselines including Cross-Stitch Networks and MTANe3
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
t0“Tomato leaf disease and severity prediction”
e1“achieves 97.85% disease classification accuracy”
e2“outperforming state-of-the-art single-task classifiers including EfficientNetV2-S, ViT-B/16, and ConvNeXt-Tiny”
e3“existing multi-task learning baselines including Cross-Stitch Networks and MTAN”
Claim · model-written summary
Ablation studies confirmed individual contributions of CBAM, MixUp and CutMix augmentation, and the cross-task attention mechanism.
- Taxon
- TomatoSolanum lycopersicumt0
- Treatment
- CBAM, MixUp and CutMix augmentation, and the cross-task attention mechanisme1
- Outcome
- individual contributionse1
Not reportedComparatorDirection
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
t0“Tomato leaf disease and severity prediction”
e1“Comprehensive ablation studies confirm the individual contributions of CBAM, MixUp and CutMix augmentation, and the cross-task attention mechanism”
Claim · model-written summary
Statistical analysis over five independent runs yielded p-values less than 0.001 for the reported improvements.
- Taxon
- TomatoSolanum lycopersicumt0
- Outcome
- p-valuese1
- Value
- less than 0.001e1
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
t0“Tomato leaf disease and severity prediction”
e1“Statistical significance analysis across five independent runs yields p-values less than 0.001”
Claim · model-written summary
Statistical analysis over five independent runs yielded Cohen's d greater than 14 for the reported improvements.
- Taxon
- TomatoSolanum lycopersicumt0
- Outcome
- Cohen's de1
- Value
- greater than 14e1
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
t0“Tomato leaf disease and severity prediction”
e1“Cohen's d greater than 14”
Claim · model-written summary
TomatoMTL achieved 89.4% Pointing Game accuracy in localizing disease regions.
- Taxon
- TomatoSolanum lycopersicumt0
- Outcome
- Pointing Game accuracye1
- Value
- 89.4%e1
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
t0“Tomato leaf disease and severity prediction”
e1“Quantitative localization analysis reveals that the model achieves 89.4% Pointing Game accuracy”
Claim · model-written summary
Attention maps focus on biologically meaningful disease regions.
- Taxon
- TomatoSolanum lycopersicumt0
- Outcome
- attention maps focus on biologically meaningful disease regionse2
- Value
- 89.4%e1
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
t0“Tomato leaf disease and severity prediction”
e1“Quantitative localization analysis reveals that the model achieves 89.4% Pointing Game accuracy”
e2“confirming that attention maps focus on biologically meaningful disease regions”