Upload a Dataset

The upload_dataset method lets you upload a dataset to a workspace into a new project or to one that already exists within Roboflow.

To upload a dataset using the Python SDK, use the following code:

import roboflow

rf = roboflow.Roboflow(api_key=YOUR_API_KEY_HERE)

# get a workspace
workspace = rf.workspace("WORKSPACE_URL")

# Upload data set to a new/existing project
workspace.upload_dataset(
    "./dataset/", # This is your dataset path
    "PROJECT_ID", # This will either create or get a dataset with the given ID
    num_workers=10,
    project_license="MIT",
    project_type="object-detection",
    batch_name=None,
    num_retries=0,
    is_prediction=False #optional, set to True if the dataset is not ground truth and needs approval
)

The num_workers value allows you to set how many images should be uploaded concurrently. We recommend setting a value no greater than 25.

To upload a dataset in the COCO JSON format, for example, make sure your dataset is structured like so:

my_dataset/
├── train/
│   ├── image1.jpg
│   ├── image2.jpg
│   ├── image3.jpg
│   └── _annotations.coco.json
├── valid/
│   ├── image4.jpg
│   └── _annotations.coco.json
└── test/
    ├── image5.jpg
    └── _annotations.coco.json