Performance Metrics for 4 Retinal Pathologies from RFMiD Dataset

1824 retinal fundus images from the Retinal Fundus Multi-Disease Image Dataset (RFMiD) were used to evaluate a lightweight CNN model. The dataset includes performance metrics such as accuracy, precision, recall, and F1-score for classifying three distinct retinal conditions, with results from an external validation on RFMiD 2.0. Author Usman Rafi published the data on figshare under a CC-BY-4.0 license.

Use Cases

• Benchmarking lightweight CNN architectures based on reported accuracy and inference time metrics.
• Studying model generalizability for retinal disease classification based on external dataset validation results.
• Analyzing the trade-off between model performance and computational efficiency based on the reported parameter count and size.

Strengths

• Includes specific performance metrics: 90.33% testing accuracy, 90.69% precision, 90.33% recall, and 90.27% F1-score.
• Model evaluation includes an external validation step using the RFMiD 2.0 dataset, suggesting a test for generalizability.
• Dataset is small (5.5 KB) and licensed permissively (CC-BY-4.0) for easy sharing and reuse.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Data may reflect bias inherent to the specific source datasets used (RFMiD and RFMiD 2.0).

Provenance

Source

figshare

Collection Method

Performance metrics derived from experiments on the Retinal Fundus Multi-Disease Image Dataset (RFMiD).

Freshness

Last updated 2026-05-11 17:32:53; freshness should be verified.