🖼️Dataset(s)

Datasets are shared between DeepCap and DeepQ AI Training, and can be used for annotation, model training and testing.

Upload Dataset: Click on this button and upload your dataset in DICOM/ JPG, PNG form.
Search: Search for your uploaded dataset by its name.
Combine: Combine two or more datasets to form a new dataset. The new dataset can be image-only or contains annotation from the source datasets.
The Dataset list shows all the datasets uploaded by the user and # of studies contained in each dataset
The number of annotation data attached is shown, each dataset can have multiple annotation data of different types of labeling and number of studies (subsets).
Users can hide the datasets without annotation by the toggle button "HIDE DATASET WITHOUT ANNOTATION DATA"
If there is no annotation data, the user has to create an annotation data via annotation project, or else it cannot be used to train an AI model.
Each dataset can contain multiple annotation data with different types or different number of studies. (the number of studies in an annotation can be a subset of the whole dataset)

Selecting a dataset brings out the detail view, showing the raw data (images) and annotation data attached under this image dataset.
Owners can delete their datasets by clicking on the option icon " ", a dialogue will pop up to double confirm deletion of the dataset and its derivative results in annotation projects, jobs, training tasks and inference jobs.
Manually key-in "DELETE" to complete the deletion process.

Deleting a dataset will remove all annotations attached to it and any derivatives on the platform, proceed with caution.

Clicking on the thumbnail of raw data(image) opens up the image viewer with essential browsing functions such as pan, zoom in/out & brightness/contrast adjustment.

Selecting any annotation file will open the annotation detail view, showing the information of the annotation file: source, annotation type and its usage across the platform.
The label distribution statistics is also provided as bar graph or table.

Clicking on each thumbnail opens the image viewer and showing the label information for a more detailed inspection.
User can toggle on/off the detection/segmentation labels by clicking on the

Last updated 2 months ago