Find Mislabeled Data Using Visual Layer Label Filters
Find mislabeled data using label filters
In many cases mislabeling can be indicated by having data labeled with multiple contradicting labels. Visual layer can surface these kind of issues. For example: a bowl that was mislabeled as a cup.
- Go to Dataset Inventoryview and select a dataset by clicking on it
- Select the required view: Images view or Objects view
- In the top filter bar, go to "Labels" filter, select one or more labels and apply the filter. Alternatively, use the right pane "Metadata" section and select one or more labels.
- The logical operator between items listed in the dropdown list of filters and within a section in the right pane is "OR".
- The logical operator between the filter types (User tags, Labels, Issues, etc.) is "AND", as well as for items selected in multiple separate sections in the right pane.

Coco Dataset: Images view, no Image labels

Coco Dataset: Objects view, Objects labels

Top bar for textual search, visual search and filters.

Coco Dataset: Top bar labels filter

Coco Dataset: Right Pane labels filter in "Metadata" section

Coco Dataset: Filter by label: "bowl"
- The returned results consist of clusters of objects that were labeled as "bowl".
- Click on the cluster in the center at the top row and then click on one of the images in the cluster to get to the "Image view".

Coco Dataset: Image view, Objects labels appear on the right. Some objects were mislabeled as cups

Coco Dataset: Image view, Bowl 5 was labeled also as cup, Cup 7
Updated 6 months ago
To select data to send for relabeling using Visual Layer continue to the next page.