52 annotation formats
Computer Vision Annotation Formats
Roboflow is the universal conversion tool for computer vision datasets. Spend more time experimenting and less time wrestling with one-off conversion scripts.
Convert Annotation Formats
799 conversion guides
/
Showing 52 of 52 annotation formats.
Featured
10- COCO JSONCOCO is a common JSON format used for machine learning because the dataset it was introduced with has become a common benchmark.JSONImportExport
- CreateML JSONApple's CreateML and Turi Create tools need a special JSON format for object detection tasks.JSONImportExport
OpenAI GPT-4o Object Detection JSONLThe format required by OpenAI for fine-tuning GPT-4o for object detection.JSONLImportExport- Pascal VOC XMLPascal VOC is a common XML annotation format that is human readable but doesn't work with any known object detection models.XMLImportExport
Tensorflow Object Detection CSVThe intermediate human-readable format prior to creating a TFRecord.CSVImportExport- Tensorflow TFRecordTensorflow TFRecords are a binary format used with the TensorFlow Object Detection models.TFRecordExport
YOLOv5 Oriented Bounding BoxesA modified version of YOLO Darknet annotations that allows for rotated bounding boxes.TXTExport- YOLOv5 PyTorch TXTA modified version of YOLO Darknet annotations that adds a YAML file for model config.TXTExport
- YOLOv8 Oriented Bounding BoxesUsed to train a YOLOv8-OBB model.TXTExport
- YOLOv8 PyTorch TXTA modified version of YOLO Darknet annotations that adds a YAML file for model config.TXTImportExport
More annotation formats
42- COCO KeypointThe keypoint detection format used in the Microsoft COCO project.JSONImportExport
COCO Run-Length Encoding (RLE)A version of the COCO JSON format with segmentation masks encoded with run-length encoding.JSONImportExport
CogniacA JSON annotation format used by Cogniac.JSONImport
Florence-2The data format used by the Florence-2 multimodal model.Export
Google Cloud AutoML Vision CSVThe CSV format needed for Google's AutoML Vision tool.CSVImportExport
IBM Cloud Annotations JSONThe native JSON format of IBM's open source Cloud Annotations labeling tool.JSONImport
Kaggle Wheat CSVCustom format used in a specific Kaggle object detection competition.CSVImport
LabelBox JSONLabelBox is an outsourced label management service, specializing in on-premise "bring your own labeler" solutions.JSONImport- LabelBox Video JSONLabelBox is an outsourced label management service, specializing in on-premise "bring your own labeler" solutions.JSONImport
- LabelMe JSONThe native format of LabelMe, an open source graphical image annotation tool written in Python and available for Windows, Mac, and Linux.JSONImport
Marmot XMLMarmot is an obscure Chinese annotation format created by researchers at Peking University.XMLImport- Multiclass Classification CSVTransforms an object detection problem into a classification problem.CSVImportExport
- Multimodal JSONLA JSONL format for multimodal datasets (i.e. VQA).JSONLImportExport
- OIDv4 TXTThe format returned by the OpenImages Dataset v4 ToolKit.TXTImport
OpenAI CLIP ClassificationCLIP is an impressive multimodal zero-shot image classifier that achieves impressive results in a wide range of domains with no fine-tuning.TXTExport- OpenImages CSVA variation on the YOLO Darknet format which removes the need for a labelmap.CSVImport
- PaliGemma JSONLPaliGemma is a multimodal model developed by Google. PaliGemma uses a JSONL data format.JSONLImportExport
RF-DETRRF-DETR is a real-time object detection transformer-based architecture designed to transfer well to both a wide variety of domains and to datasets big and small.JSONImportExport
RetinaNet Keras CSVA custom format used by the Keras implementation of RetinaNet.CSVImportExport
Sagemaker GroundTruth ManifestThe proprietary annotation format used by Amazon SageMaker GroundTruth.JSONImportExport- Scale AI JSONScale AI is a popular outsourced labeling platform; their API returns object detection annotations in this JSON format.JSONImport
- Scaled-YOLOv4 TXTScaled-YOLOv4 uses a variant on the Darknet TXT format with an additional data.yaml configuration file.TXTExport
Segment Anything 2SAM-2 uses a custom dataset format for use in fine-tuning models. You can convert and export data to the SAM 2 format in Roboflow.JSONExport
SuperAnnotate JSONSuperAnnotate (formerly annotate.online) is a self-service labeling tool and outsourced annotation provider.JSONImport- Supervisely JSONSupervise.ly offers a computer vision platform and uses a proprietary JSON-based annotation format.JSONImport
Udacity TXTThe format used by the Udacity Self Driving Car dataset.TXTImport
Unity Perception JSONUnity Perception is a toolkit for generating datasets for computer vision by using 3D renders.JSONImport
VGG Image Annotator CSVThe VGG Image Annotator tool's CSV format.CSVImport- VGG Image Annotator JSONThe VGG Image Annotator tool's JSON format.JSONImport
VoTT CSVThe native format of Microsoft's Visual Object Tagging Tool (VoTT)CSVImport- VoTT JSONThe native format of Microsoft's Visual Object Tagging Tool (VoTT)JSONImport
YOLOYOLO PyTorch TXT is a format commonly used by YOLO vision models like YOLO26, YOLO11, YOLOv8, and YOLOv5.TXTImportExport- YOLO Darknet TXTThe favored annotation format of the Darknet family of models.TXTImportExport
- YOLO Keras TXTA format used with the Keras implementation of YOLOv3.TXTImportExport
YOLOv10 PyTorch TXTA modified version of YOLO Darknet annotations that adds a YAML file for model config.TXTImportExport- YOLOv11 PyTorch TXTTXTImportExport
- YOLOv12 PyTorch TXTTXTImportExport
- YOLOv4 PyTorch TXTA format used with the PyTorch port of YOLO v4.TXTExport
- YOLOv7 PyTorch TXTA modified version of YOLO Darknet annotations that adds a YAML file for model config.TXTExport
- YOLOv8 Keypoint TXTThe keypoint format used by YOLOv8.TXTImportExport
- YOLOv9 PyTorch TXTA modified version of YOLO Darknet annotations that adds a YAML file for model config.TXTImportExport
- meituan/yolov6MT-YOLOv6 uses a modified version of YOLO Darknet annotations.TXTExport