Ablation Study Results for Cocoa Pod Disease Detection Models

98.95% accuracy was achieved by a novel LGF-CBAM deep learning model on the Cocoa_Pod_Disease_Gh dataset. This 5.5 KB XLS file contains results from an ablation study for activation function selection, authored by Henry Techie-Menson and last updated in April 2026. The study cross-validated the model on four additional cocoa disease datasets, with reported accuracies ranging from 94.00% to 98.53%.

Use Cases

• Benchmarking new attention mechanisms for image classification based on the described LGF-CBAM architecture.
• Comparing model performance across different agricultural disease datasets based on the cross-dataset evaluation results.
• Analyzing the impact of activation functions on model accuracy for precision agriculture applications.

Strengths

• Model achieved a high accuracy of 98.95% on the primary dataset.
• Cross-dataset evaluation demonstrates robustness across four distinct cocoa disease image collections.
• Dataset is openly licensed under CC-BY-4.0.

Limitations

• Dataset size is only 5.5 KB, indicating very limited scope.
• Row count and specific column-level documentation are unknown.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Henry Techie-Menson via figshare

Collection Method

Likely contains tabular results from a machine learning ablation study.

Freshness

Last updated 2026-04-30 17:51:56; freshness should be verified.