Resultsets

A resultset is a collection of curated data objects the user has selected through different explore and refine capabilities provided by Data Explorer. Using an adectl command, the user can upload the resultset objects into the S3 bucket, Azure blob store, or a local file system.

Create a Resultset for Image-based job

1. Navigate to Repo > Dataset, and select the dataset that you want to query and save as a resultset.

   

2. On the Dataset card, click Visualize.

   

3. On the dataset screen, click one of the cluster points to view the frames.

   

4. Click the + (plus) icon at the bottom-right corner to save the frame results as a resultset, for future reference.

   

   You can select the top 60 image frames(or all those images that you have scrolled through), all 100 frames, or the cluster group to save as a resultset.

5. Save the results, by either adding to an existing resultset or creating a new resultset.

   

   1. Click Add to existing to add the results to an existing resultset, if there is already one.

      1. In the Select Resultset window, enter the name of an existing resultset

      2. Click the Select button.

         

   2. Click Create new, if you want to create a new resultset

      1. In the Create Resultset window, enter Name and Description

      2. Click the Submit button.

         

The result set gets created successfully. 

The resultset listing page, as shown below, provides a list of all resultset created. This page provides sorting, filtering, and search capabilities to reach the specific resultset of interest.

Add Objects to the Resultset

• Click the '+' icon against each thumbnail to add the image frame to the resultset individually. 

• Click the '+' icon at the button-right corner for bulk addition of the image frames to the resultset. 

If there is no resultset in an opened state (when adding to the resultset for the first time), a form to enter the name of the resultset will be presented.

You can also search for video sequences and save as resultset.

Create Resultset for Video Sequence Job

1. Navigate to Repo > Dataset, and click the Visualize button on the job card for the video sequence dataset.
   

2. Click a data point to view all image frames with associated video sequence, select an image, and click the thumbs-up icon.

3. Click an image frame and enable the Patch option.
   
   By default, the Trajectory mode is enabled for the selected frame for which the patch pipeline is already run on the video sequence.
   

4. Select the Frames icon on the selected images to open the series of image frames.
   

5. Select the key frames to mark the objects and define the trajectory path.
   

   1. Draw the bounding box to mark the same object in a specific direction, say from Left to Right, in the frames.

   2. Click OK.

6. Click Search on the dataset page to search for the frames based on the defined trajectory mode.
   

7. From the search results, select an image frame. and click the Frames icon to view the results with the matching trajectory as per the key frames.
   

   The page displays the number of matched keyframes.
   

8. Enable the options Show Matching Frames Only and Show Matching Regions.
   As you can see that the image frames displays the matching regions as per the trajectory you had defined with the bounding box for search.
   You can review each such frame from the search results.

9. Once you are satisfied with the search results, click the + (plus) sign at the bottom-right corner of the page, and select all points to save the results as a result set.
   

You can save the result to an existing resultset or create a new result set as explained under the Create a ResultSet section.

Open an Existing Resultset

In the resultset tab, click the Select a Resultset drop-down arrow as shown in the below image and choose a resultset,

Explore Resultset Actions

The ResultSet page provides information about the image objects, displays the ID for the resultset, and allows you to perform various actions.

• Existing Images: Represents the total number of objects in the resultset.

• To Be Added Images: Represents the number of objects that are added to the resultset but not yet saved.

• To Be Deleted Images: Represents the number of objects that are deleted from the resultset but not yet saved.

On the Resultset page, you can perform the following actions.

• Rename resultset: Click the Rename button on the top to rename the resultset, as needed.
  

  

• Close resultset: Click the Close button to close the resultset.

• Refresh resultset: Click the Refresh icon on the top-right corner to refresh and add or remove you had marked any image frames for addition or deletion.

• Delete resultset: Click the Delete icon to delete the Resultset. This action cannot be undone.

• Download resultset: Click the download icon to download the resultset locally, in JSON or CSV format, as needed.
  

  

Publish a resultset

You can publish a resultset for all users to access it outside the job.

1. On the resultset page, click the Publish button available on the top right.

2. Click Publish on the confirmation message window.  

   

A resultset can be published for access outside the job context. The published resultsets are available on the 'Resultsets' page, accessible from the left panel, as shown below.

Note

A publish operation cannot be reverted.

Any edits to the resultset after publishing will reflect in the resultset when accessed from the Resultset listing page.

• Save and Save As: If you make any changes to the resultset, such as deleted a few images, or renamed the resultset, you can save it, as needed.

Resultset Upload

The resultset upload operation allows the user to materialize the objects in a resultset onto a target S3 bucket, Azure blob store, or a local file system for further analysis or connecting to downstream pipelines (e.g., training pipeline).

Configure the Target Location

adectl resultset config

This command will select the type of destination location (S3/Azure/GCP. etc.) and necessary credentials and configuration information. As an example, the fields captured for the S3 destination are as follows:

Select store type [s3 | azure | GCP | file | hdfs] : s3
Enter S3 Bucket Name: bucket
Enter S3 Access Key: xxxxx
Enter S3 Secret Key: yyyyy
Enter S3 Endpoint [default: https://s3.amazonaws.com]: 
Configured S3 Store Successfully

 For Azure blob store as a destination, storage account key fields are needed.

Upload resultset objects

You can upload resultsets using either resultset ID or resultset name. 

adectl resultset upload -n <resultsetname> -t <target-location>

OR

adectl resultset upload -r <resultsetid> -t <target-location> 

The command starts the upload operation as an asynchronous operation, and the status of this operation is available by running the following command.

adectl show

Note

If the resultset name is not unique, the command will prompt you to invoke the adectl resultset upload command with resultsetid.

Dump Resultset Objects

The command copies all the object file names available in the selected resultset.

adectl resultset dump -n <resultsetname> -o <output-dir>

OR

adectl resultset dump -r <resultsetid> -o <output-dir>