How darkanalytica1/drone-detection-eval works
ddeval is a dependency-free evaluation harness for small-drone detection and tracking, not a detector. The main workflow accepts MOT or Anti-UAV ground truth plus MOT predictions, evaluates them at a supplied frame rate, and prints or writes a report. Its demo generates a synthetic scene and simulated detections before running the same evaluation. The graph preserves input handling, matching, metric calculation, reporting, and the documented demo capability; edges reflect calls or data use visible in the excerpts.
Main entry points: __main__.py CLI commands · metrics.py Metrics and tracks · formats.py Input and MOT formats · matching.py Box matching
Components 8
Command workflow
- CLI commands
ddeval/__main__.py - Input and MOT formats
ddeval/formats.py
Evaluation core
- Metrics and tracks
ddeval/metrics.py - Box matching
ddeval/matching.py - Text and Markdown reports
ddeval/report.py
Synthetic demo
- Synthetic scene
ddeval/synth.py - Simulated detector
ddeval/synth.py
Other components
- User
Connections 7
- User → CLI commands: runs
README.md - CLI commands → Input and MOT formats: reads or writes
ddeval/__main__.py - CLI commands → Metrics and tracks: evaluates
ddeval/__main__.py - Metrics and tracks → Box matching: matches boxes
ddeval/metrics.py - CLI commands → Text and Markdown reports: formats results
ddeval/__main__.py - CLI commands → Synthetic scene: generates demo
ddeval/__main__.py - CLI commands → Simulated detector: simulates predictions
ddeval/__main__.py