Images

Go to Images to upload data, assign images, run smart segmentation and prediction, and more.

Upload pre-annotation

Use the Upload pre-annotation function when you have completed a certain amount of predictions for a project or when you have started a project on a different platform and want to migrate your work without any data loss. Learn how to upload pre-annotations here.

Upload data

Pixel projects

Select Upload and choose one of the four ways to upload images. You can upload up to 2000 images at a time. We recommend you upload up to 1000 images at a time for a faster and smoother performance.

Upload from computer

Drag and drop or choose images from your computer. SuperAnnotate supports images in the following formats: .jpg, .jpeg, .png, .webp,.tiff, and .bmp.

Note that two images cannot have the same name. An image that has a name similar to an existing image will be skipped.

Consider using the following methods when uploading many images:

Import from S3 bucket

Enter the access key ID, secret access key, bucket name, and folder name (optional). Select Test to see whether you have access to the S3 bucket or not. If you have access, go ahead and click Start.

CLI upload

Use the command line uploader when you want to upload a large batch of images. Learn more here.

Python SDK

Upload images using Python SDK. To do that, select Upload Annotation, choose a format, and use the displayed command. Learn more here.

Vector projects

Image

Select Upload > Image, and choose one of the four ways to upload images. You can upload up to 2000 images at a time. We recommend you upload up to 1000 images at a time for a faster and smoother performance. Note that free users can upload up to 100 images only.

Upload from computer

Drag and drop or choose images from your computer. SuperAnnotate supports images in the following formats: .jpg, .jpeg, .png, .webp,.tiff, and .bmp.

Note that two images cannot have the same name. An image that has a name similar to an existing image will be skipped.

Consider using these methods when uploading many images.

Import from S3 bucket

Enter the access key ID, secret access key, bucket name, and folder name (optional). Select Test to see whether you have access to the S3 bucket or not. If you have access, click Start.

CLI upload

Use the command line uploader when you want to upload a large batch of images. Learn more here.

Video

Select Upload > Video, and drag and drop or choose videos from your computer. SuperAnnotate supports videos in the following formats: .mp4, .avi, .mov, .flv, .mpeg, and .WebM.

Under Frame rate (fps), enter a number to break your videos into several frames or images. Frame rate (fps) is the number of frames that appear per second. The higher the frame rate, the more images will appear per second, and vice versa. The default frame rate is 1.

Request images

Annotators and QAs can request images by selecting Request images in the Images tab or by clicking the plus button in the Image panel in the editor. They will be automatically assigned up to 50 images and cannot have more than 50 images at a time. Annotators and QAs have to finish working on the assigned images to be able to request more images.

Run smart prediction

Use smart prediction to cut down your annotation time. Select an image or multiple images, select Run smart prediction, choose a model, and start the prediction. When the smart prediction is completed, a green tick will appear on the smart prediction icon.

Run smart prediction button

Types of models

For pixel projects:

  • Instance segmentation (trained on COCO): Predicts your instance’s class label, bounding box, and binary mask. Use this model to detect countable objects, such as cats.

  • Panoptic segmentation (trained on COCO): Assigns all the pixels to classes. Use this model to detect countable objects, such as cars, and uncountable objects, such as the sky.

For vector projects:

  • Cars: Predicts vehicles.

  • Instance segmentation (trained on COCO): Predicts your instance’s class label, bounding box, and binary mask. Use this model to detect countable objects, such as cats.

  • Human pose keypoint detection (trained on COCO): Predicts human beings.

  • OCR (English text): Predicts texts in English.

  • Building detection for aerial imagery: Predicts buildings.

  • Object detection (trained on COCO): Predict objects using bounding boxes.

Run smart segmentation

Run smart segmentation on your images to ease the annotation process. Smart segmentation works only on pixel projects. Select an image or multiple images and select Run smart segmentation.

Run smart segmentation button

Next, choose a model:

  • Generic

  • Autonomous driving

After choosing the model, select Start segmentation.

When the smart segmentation is completed, a green tick will appear on the smart segmentation icon. The editor will now be able to divide your image into multiple segments.

Upload entropy values (Active Learning)

The entropy value feature helps you determine the images that you need to annotate first to improve your model training. It’s a key Active Learning feature that speeds up your annotation time.

To upload entropy values, select the icon that reads CSV. Next, drag and drop or choose a CSV file from your computer, and select Upload. The CSV file is generated by the user in advance. It consists of a table that stores image names and their numeric values.

After the upload is completed, an entropy value appears next to each image. You can now sort your images by their entropy values.

Images that don’t have names or entropy values will not be uploaded.

Assign images

To assign images to your project’s contributors, select an image or multiple images and select the person icon.

Assign images button

Choose Annotation or QA, depending on what type of work needs to be done on your image(s) and select a user from the drop-down menu. Remember, you can assign images only to Annotators and QAs.

To remove an assignment, select an image or multiple images and select the person icon. Choose Annotation or QA and select Remove assignments. Your image(s) will be unassigned.

Sort images

Select the drop down arrow, and sort the images by:

  • Image name

  • Prediction: Not started, In progress, Completed, and Failed.

  • Segmentation: Not started, In progress, Completed, and Failed.

  • Status: Not started, In progress, Pending, Quality Check, Completed, and Skipped.

  • Entropy

To find an image, select the search icon and enter its name.

Filter images

Select the filter icon and filter the images based on the:

  • Assignee

  • Annotation status: Not started, In progress, Quality check, and Returned.

  • Prediction: Not started, In progress, Completed, and Failed.

  • Segmentation: Not started, In progress, Completed, and Failed.

To filter pinned images, select the pin icon > Apply.

Delete images

You can delete images one by one or in bulk.

To delete an image, check the box next to the image and select Delete. Only Team owners, Team admins, Project admins (with permission), and customers can delete images.

To delete images in bulk, check the box next to Image to select all the images displayed on the page, and select Delete. Only Team owners, Team admins, and Project admins (with permission) can delete images in bulk.