Get Started

Run Your First Evaluation

Score the sealed Prediction Set with Performance and a minimal Robustness configuration.

The Evaluation scores an existing Prediction Set. It does not rerun or modify the predictions you generated in the previous step.

Configure the Evaluation

  1. From the Prediction Set detail, choose Evaluate.
  2. Keep Manual as the Evaluation configuration.
  3. Choose Configure and select Detection mAP 50-95 as the Performance metric.
  4. Open the Robustness tab and include Fog.
  5. Select Fog severity 3, matching the Prediction Set coverage.
  6. Set Maximum allowed degradation to 20%.
  7. Choose Apply changes and confirm that readiness becomes Ready.

Manual Evaluation with Detection mAP 50-95, Fog severity 3, a 20 percent degradation limit, and Ready state.

Watch: Configure an EvaluationSelect Detection mAP 50-95, Fog, and a 20% degradation criterion. Launch the configured run after reviewing readiness.Silent demo · Subtitles follow page language · Full screen for detail

Launch and wait for completion

  1. Choose Continue.
  2. Name the Evaluation Helmet 100 robustness evaluation.
  3. Choose Launch.
  4. Follow progress until the Evaluation detail shows Completed.

Completed Evaluation Performance result showing Detection mAP 50-95 and 100 validation samples.

Watch: Read completed resultsRead the expanded Fog comparison and its 20% criterion.Silent demo · Subtitles follow page language · Full screen for detail

Completed means the required computation finished and selected optional branches completed or recorded an allowed degraded outcome. It does not mean that every metric passed its threshold.

Continue to Inspect the Claim and Evidence.