Dataset Upload: Roboflow Data Types

An overview of the data formats available for upload to Roboflow.

Written by Mohamed Traore

Last published at: September 28th, 2022

How to successfully upload your images, or videos, into Roboflow.

Data can currently be added to Object Detection, Single-Label Classification, Multi-Label Classification, Instance Segmentation, and Semantic Segmentation projects (semantic segmentation: upload of masks is not yet supported, only annotations and images - export of masks is supported). For interest adding data for other project types requesting early access is highly encouraged.

What data can I upload?

Roboflow can ingest:

  1. Various image formats, including JPG, PNG, and BMP.
  2. Video formats, including MOV, MP4, and AVI.
  3. Various object detection, classification and segmentation annotation formats, including JSON, XML, CSV, and TXT.
  4. Data from Roboflow Universe!
  5. Object Detection vs. Image Classification vs. Keypoint Detection: What's the Difference?

Roboflow's Dataset Ownership Policy: Who owns my data?

  • We strongly believe that your images and videos are yours. That's why, when you upload those, your photos and your videos remain yours -- we do not own them. (You can check out additional details in our terms of service, item 22B.)

How can I add data?

You can add data to your Roboflow account by:

  1. Using the Web User Interface (Recommended) - Uploading Object Detection and Instance Segmentation Datasets || Uploading Classification Datasets
  2. Using our API - Obtaining Your API Key
  3. Forking/downloading a publicly available image dataset from Roboflow Universe.

Uploading images with existing annotations:

Roboflow ingests over 25 different annotation formats for computer vision.

After import, Roboflow can convert (export) annotated datasets in over 20 annotation formats.

More on: Workspaces, Projects, and Versions

  • includes: how Workspaces, Projects, and dataset/project Versions work on Roboflow, and notes on when each project type is useful (Object Detection, Classification, Segmentation, Keypoint).

The Labeling Interface

  • How it works: making sense of the User Interface (UI) and Shortcut Keys