Export Data

Export data for use in model training.

You can export a dataset from the Roboflow platform using the web interface and the Python SDK.

Python SDK

To create a ZIP file of a dataset for export from the Python SDK, use the project.export() method. To download the ZIP file, use the project.download() method.

version.download(model_format="yolov5", location="./downloads")

REST API

/:workspace/:project/:version/:format is the route you should use to get the download link for an exported dataset in a specific format. You can use this in the Jupyter notebooks from our model library or your own custom training scripts.

The following endpoint returns an export value that contains a link key with a URL from which you can download a dataset:

curl "https://api.roboflow.com/roboflow/chess-sample-4ckfl/1/yolov5pytorch?api_key=$ROBOFLOW_API_KEY"

Here is an example payload returned by the endpoint:

{
    "workspace": {
        "name": "Roboflow",
        "url": "roboflow",
        "members": 7
    },
    "project": {
        "id": "roboflow/chess-sample-4ckfl",
        "type": "object-detection",
        "name": "Chess Sample",
        "created": 1630335544.592,
        "updated": 1630335741.988,
        "images": 12,
        "unannotated": 3,
        "annotation": "pieces",
        "public": false,
        "splits": {
            "test": 1,
            "train": 9,
            "valid": 2
        },
        "classes": {
            "white-bishop": 11,
            "black-king": 8,
            "black-knight": 11,
            "white-queen": 7,
            "black-bishop": 8,
            "white-rook": 10,
            "black-rook": 10,
            "white-king": 8,
            "black-queen": 4,
            "black-pawn": 37,
            "white-pawn": 34,
            "white-knight": 10
        }
    },
    "version": {
        "id": "roboflow/chess-sample-4ckfl/1",
        "name": "augmented",
        "created": 1630335698.746,
        "images": 30,
        "splits": {
            "train": 27,
            "test": 1,
            "valid": 2
        },
        "model": {
            "id": "chess-sample-4ckfl/1",
            "endpoint": "https://detect.staging.roboflow.com/chess-sample-4ckfl/1",
            "start": 1630335799.682,
            "end": 1630337523.889,
            "fromScratch": false,
            "tfjs": true,
            "oak": true,
            "map": "62.87",
            "recall": "85.29",
            "precision": "23.44"
        },
        "preprocessing": {
            "grayscale": {
                "enabled": true
            },
            "resize": {
                "width": 416,
                "height": 416,
                "enabled": true,
                "format": "Stretch to"
            },
            "auto-orient": {
                "enabled": true
            }
        },
        "augmentation": {
            "rotate": {
                "enabled": true,
                "degrees": "5"
            },
            "exposure": {
                "enabled": true,
                "percent": "25"
            },
            "noise": {
                "enabled": true,
                "percent": "2"
            },
            "image": {
                "versions": "3",
                "enabled": true
            },
            "flip": {
                "horizontal": true,
                "enabled": true,
                "vertical": false
            },
            "brightness": {
                "enabled": true,
                "brighten": true,
                "percent": "25",
                "darken": true
            },
            "crop": {
                "percent": 30,
                "enabled": true,
                "min": 0
            }
        },
        "exports": [
            "yolov5pytorch"
        ]
    },
    "export": {
        "format": "yolov5pytorch",
        "link": "https://app.staging.roboflow.com/ds/o06EL6FAt7?key=KYtxHKCvaN"
    }
}

CLI

You can export data through the Roboflow CLI using the following command:

roboflow download <datasetUrl>

Find more information about this CLI command in our docs.

Here are the settings options available for dataset export:

Object Detection	Single-Label Classification	Multi-Label Classification	Instance Segmentation	Semantic Segmentation	Keypoint Detection
clip	folder	multiclass	coco-segmentation	coco-segmentation	coco
coco	clip	folder	clip	png-mask-semantic	yolov5pytorch
createml		clip	coco
darknet			createml
multiclass			darknet
tensorflow			multiclass
tfrecord			tensorflow
voc			tfrecord
yolokeras			voc
yolov5pytorch			yolokeras
yolov7pytorch			yolov4pytorch
mt-yolov6			yolov4scaled
retinanet			yolov5-obb
benchmarker			yolov5pytorch
			yolov7pytorch
			mt-yolov6
			retinanet
			benchmarker