SuperAnnotate is a provider of outsourced labeling annotation services and develops a self-serve labeling tool (formerly known as annotate.online) which is available for download on Windows, Linux, and Mac. Their tool has support for advanced labeling functions like video frame annotation and using a trained model to pre-suggest annotations.
You can use Roboflow to convert SuperAnnotate's custom JSON format to any format your computer vision models expect.
Below, learn the structure of SuperAnnotate JSON.
{
"mask.jpg": [{
"type": "bbox",
"classId": 1,
"probability": 100,
"points": {
"x1": 67.8,
"x2": 187.6,
"y1": 166.6,
"y2": 274.1
},
"groupId": 0,
"pointLabels": {},
"locked": false,
"visible": true,
"attributes": []
}, {
"type": "polyline",
"classId": 2,
"probability": 100,
"points": [148.7, 321.1, 164.5, 328.3, 220.4, 335.5, 280.6, 344.8, 330.1, 341.9, 373.1, 330.5, 375.3, 336.2, 377.4, 369.9, 382.4, 384.9, 377.4, 409.3, 360.2, 425.1, 318.6, 451.6, 279.9, 473.8, 262.7, 488.2, 252, 499.6, 241.9, 502.5, 223.3, 496.8, 210.4, 486, 181.7, 453.8, 165.9, 443, 155.9, 431.5, 148, 419.4, 132.3, 410, 123.7, 397.8, 139.4, 346.2, 151.6, 319.7, 150.9, 321.1],
"groupId": 0,
"pointLabels": {},
"locked": false,
"visible": true,
"attributes": []
}, {
"type": "polygon",
"classId": 2,
"probability": 100,
"points": [493.7, 355.6, 506.1, 366.1, 567.3, 384.2, 648.5, 352.7, 649.5, 372.8, 659, 386.1, 649.5, 402.4, 631.3, 424.4, 618.9, 451.1, 612.2, 456.9, 585.4, 474.1, 570.1, 477.9, 552, 466.4, 527.1, 454, 491.8, 402.4, 494.6, 377.5, 493.7, 355.6],
"groupId": 0,
"pointLabels": {},
"locked": false,
"visible": true,
"attributes": []
}, {
"type": "meta",
"name": "lastAction",
"timestamp": 1600628979793
}]
}[{
"attribute_groups": [],
"color": "#32a852",
"id": 1,
"name": "no-mask",
"opened": true
}, {
"attribute_groups": [],
"color": "#d6d619",
"id": 2,
"name": "mask",
"opened": true
}]With Roboflow supervision, an open source Python package with utilities for completing computer vision tasks, you can merge and split detections in SuperAnnotate JSON. Read our dedicated guides to learn how to merge and split SuperAnnotate JSON detections.
Below, see model architectures that require data in the SuperAnnotate JSON format when training a new model.
On each page below, you can find links to our guides that show how to plot predictions from the model, and complete other common tasks like detecting small objects with the model.