Proposal: Add standardized eye-tracking benchmark evaluation endpoint

[Sahil-aka]

Hi,

I implemented a structured benchmarking module for evaluating eye-tracking accuracy and precision, aligned with the project’s validation goals.

Key additions:

• Accuracy metrics (mean, median, p95) in pixels and degrees
• Scientifically grounded per-target RMS precision metric
• Data quality reporting
• Per-target accuracy breakdown
• New API endpoint: POST /api/session/benchmark
• Input validation and NumPy serialization handling

This enables reproducible and standardized benchmarking across devices and setups.

Before opening a PR, I’d like feedback on:

• Metric definitions
• API structure
• Naming conventions
• Integration approach

Looking forward to your thoughts.

[Ash6604]

Hii, looks like a great implementation

I would like to contribute with by extending the benchmark evaluation:
1.Visualization - Gaze scatter plots, heatmap for better interpretability.
2.Additional error analysis - distribution plots to understand the error spread and outliers.
3.Export Functionality - CSV Outputs for research use.

I believe these additions would complement the existing metrics and make the benchmark evaluation module more useful.
Please let me know whether this direction aligns with this project and I would be happy working on this issue with these enhancements.