DSE-YOLO11: Model Configuration and Performance Metrics for Traffic Object Detection

DSE-YOLO11 improves recall from 0.744 to 0.811 and mAP50 from 0.810 to 0.856 on the RAD dataset, while maintaining 2.96M parameters and 7.1 GFLOPs. The dataset, authored by Yange Chen and last updated in June 2026, contains configuration details for this lightweight adaptation of YOLO11n designed for detecting traffic elements in complex road scenes. It is a 5.5 KB XLS file shared under a CC-BY-4.0 license on figshare.

Use Cases

• Benchmarking lightweight object detection models based on reported recall and mAP50 metrics
• Analyzing the impact of DynamicConv, SlimNeck, and EMA integration strategies on detection performance
• Evaluating model efficiency for real-time applications based on parameter count and GFLOPs
• Comparing detection robustness for small, low-contrast traffic elements in complex scenes

Strengths

• Quantifies performance gains with specific metrics: recall improved to 0.811 and mAP50 to 0.856
• Reports model efficiency metrics: 2.96M parameters and 7.1 GFLOPs
• Includes preliminary external validation on the BDD100K dataset

Limitations

• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment
• The description suggests broader validation is still needed for external datasets

Provenance

Source

Yange Chen

Collection Method

Experiments conducted on the RAD dataset; results from model integration strategy.

Freshness

Last updated 2026-06-01 17:34:25