SSH-YOLO: Object Detection Model Performance Comparison Results

5.5 KB of tabular results comparing the SSH-YOLO model's performance against YOLOv8n on multiple datasets. The data, authored by Tenglong Ma and last updated in April 2026, contains evaluation metrics from experiments on a self-built RoadScene-Complex dataset and public datasets including BDD100K, KITTI, COCO, and CityPersons. It supports a research paper proposing an improved model for detecting dense, small, and occluded objects in complex traffic scenarios.

Use Cases

• Benchmarking object detection model accuracy based on reported [email protected] scores across multiple datasets.
• Evaluating model performance on small targets (< 32x32 pixels) based on the described four-level detection system.
• Comparing inference speed and lightweight characteristics for real-time applications based on the 60 FPS claim.
• Analyzing detection improvements in occluded scenes based on results from the CityPersons dataset.

Strengths

• Provides specific performance metrics, including [email protected] improvements of 12.4% on a self-built dataset and 22.8% on an occluded scene dataset.
• Results are derived from experiments on five distinct datasets, suggesting a multi-faceted evaluation.
• File size is 5.5 KB, indicating a compact and focused set of comparison data.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for statistical analysis.
• The dataset's scope is limited to the specific model comparison described; it is not a primary image dataset.

Provenance

Source

Tenglong Ma via figshare.

Collection Method

Likely contains experimental results from training and evaluating the SSH-YOLO model.

Freshness

Last updated 2026-04-15 18:05:46; freshness should be verified.