Hyperparameter Effects on LiDAR Segmentation Model Loss

5.5 KB of tabular data from Xuewen Feng, published in April 2026, analyzes hyperparameter impacts on loss functions for LiDAR-based semantic segmentation models. The data supports research on the MSCNet architecture, which was validated on SemanticKITTI, SemanticPOSS, and Pandaset benchmarks. It documents experiments balancing model parameters, accuracy, and computational speed for real-time robotics applications.

Use Cases

• Analyze loss function sensitivity to specific hyperparameters for the MSCNet model.
• Compare segmentation performance metrics across different hyperparameter configurations.
• Validate the trade-off between model parameter count and segmentation accuracy reported in the paper.

Strengths

• Data is associated with a model validated on three public benchmarks: SemanticKITTI, SemanticPOSS, and Pandaset.
• Released under a permissive CC-BY-4.0 license for open use.

Limitations

• The dataset is extremely small at 5.5 KB, indicating limited experimental scope or summary-level results only.
• No sample data or column descriptions are provided, obscuring the specific variables and metrics captured.

Provenance

Source

Xuewen Feng via figshare.

Collection Method

Likely experimental results from training and evaluating the MSCNet semantic segmentation model.

Freshness

Last updated in April 2026.