fiftyone-model-evaluation

What problem does it solve?

This Skill enables you to evaluate model predictions against ground truth across multiple evaluation protocols (COCO, Open Images, and custom methods) within FiftyOne.

Core Features & Use Cases

• Interactive Evaluation: Run COCO, Open Images, or custom metric evaluations from the Model Evaluation Panel.
• Programmatic Evaluation: Use the Python SDK to perform detections, classifications, segmentations, and regressions with configurable keys and metrics.
• Real-world Scenario: Compare two object-detection models by computing mAP, precision, recall, and per-class metrics, then inspect failures via evaluation patches.

Quick Start

• set_context(dataset_name="my-dataset")
• dataset_summary(name="my-dataset")
• launch_app(dataset_name="my-dataset")
• execute_operator( operator_uri="@voxel51/evaluation/evaluate_model", params={ "pred_field": "predictions", "gt_field": "ground_truth", "eval_key": "eval", "method": "coco", "iou": 0.5, "compute_mAP": true } )