Data Quality#
Scan a FiftyOne Enterprise dataset for image quality issues — including blur, brightness anomalies, low entropy, extreme aspect ratios, and duplicates — and review them interactively in the Data Quality panel.
Requirements#
Usage#
Ask your AI assistant:
"Check if my training data has bad images"
"Find all blurry or overexposed images in my dataset"
"Are there duplicates in my dataset?"
"Clean my dataset before training"
The skill audits existing quality fields, computes missing metrics as background jobs, presents a quality report with flagged counts per issue type, and opens the Data Quality panel for threshold-based interactive review and tagging.