Computer Vision Model Performance Comparison: Hybrid, ResNet50, and Swin Transformer V2

5.5 KB of tabular data comparing the performance of a proposed hybrid neural architecture with and without an ECA module against baseline models ResNet50 and Swin Transformer V2. The dataset, authored by Christopher Mai, is available under a CC-BY-4.0 license and was last updated on May 29, 2026. It is stored in an XLS file format.

Use Cases

• Benchmarking neural network architectures based on the described performance comparison.
• Analyzing the effect of Efficient Channel Attention (ECA) modules on a hybrid model's performance.
• Comparing the performance of convolutional (ResNet50) and transformer-based (Swin Transformer V2) baselines.

Strengths

• Dataset is small (5.5 KB), facilitating quick download and inspection.
• Uses a permissive CC-BY-4.0 license, allowing for broad reuse and modification.
• Provides a direct comparison of multiple modern computer vision architectures.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The description is minimal, offering no details on evaluation metrics, tasks, or datasets used for the comparison.

Provenance

Source

figshare

Freshness

Last updated 2026-05-29 17:30:05; freshness should be verified.