Skip to main content
Ctrl+K
Datumaro 1.10.0 documentation - Home
  • Docs
  • GitHub
  • Docs
  • GitHub

Section Navigation

Get Started

  • Introduction
  • Quick Start Guide
    • Installation
    • Usage
    • Examples

Level Up

  • Basic Skills
    • Level 1: Project Manipulation
    • Level 2: Dataset download
    • Level 3: Data Import and Export
    • Level 4: Detect Data Format from an Unknown Dataset
  • Intermediate Skills
    • Level 5: Data Subset Aggregation
    • Level 6: Data Comparison with Two Heterogeneous Datasets
    • Level 7: Merge Two Heterogeneous Datasets
    • Level 8: Dataset Validation
    • Level 9: Dataset Filtering
    • Level 10: Dataset Explorartion from a Query Image/Text
    • Level 11: Data Generation
  • Advanced Skills
    • Level 12: Project Versioning
    • Level 13: Pseudo Label Generation
    • Level 14: Dataset Pruning

Data Formats

  • Supported Data Formats
    • ADE20k (v2017)
    • ADE20k (v2020)
    • Align CelebA
    • Arrow
    • AVA Action
    • BraTS
    • BraTS Numpy
    • CelebA
    • CIFAR
    • Cityscapes
    • COCO
    • Common Semantic Segmentation
    • Common Super Resolution
    • CVAT
    • DatumaroBinary
    • Datumaro
    • DOTA
    • ICDAR
    • Image zip
    • ImageNet
    • Kaggle
    • Kinetics
    • KITTI
    • Velodyne Points / KITTI Raw 3D
    • LabelMe
    • LFW
    • Mapillary Vistas
    • Market-1501
    • MARS
    • MMDetection COCO
    • MNIST
    • Multiple Object Tracking (MOT)
    • Multiple Object Tracking and Segmentation (MOTS)
    • MPII Human Pose JSON
    • MPII Human Pose
    • MVTec AD
    • NYU Depth Dataset V2
    • Open Images
    • Pascal VOC
    • Roboflow
    • Segment Anything
    • Supervisely Point Cloud
    • SYNTHIA
    • Tabular
    • Vgg Face2 CSV
    • Video
    • VoTT CSV
    • VoTT JSON
    • WIDER Face
    • YOLO
    • YOLO-Ultralytics
  • Supported Media Formats
  • Datumaro Format

Hands-on Examples

  • Dataset I/O
    • Import and Export Public Data
    • Import and Export Public Semantic Segmentation Data
    • Encrypt Your Dataset
  • Manipulate
    • Merge Multiple Datasets for Classification Tasks
    • Merge Heterogeneous Datasets for Detection
  • Explore
    • Visualize Your Data
    • Find Most Similar Data from Image or Text Queries
  • Refine
    • Validate Dataset To Detect Anomalies
    • Correct Dataset from Validation Report
    • Filter Data through Your Query
    • Prune Dataset to extract representative subset
  • Transform
    • Transform Dataset: Re-id, Reindexing, Remapping, etc.
    • Tile your Dataset to Cope with High-Resolution Images
    • Transform bounding box to instance mask annotations using Segment Anything Model
    • Automatic instance mask annotations generation using Segment Anything Model
  • From Datumaro to Model Training
    • Train Your OpenVINO™ Model Using YoloV8 Trainer For Any Dataset Format
    • Noisy label detection for classification tasks from training loss dynamics statistics
    • Noisy label detection for detection tasks from training loss dynamics statistics
    • Missing Annotation Detection

Command Line Reference

  • Overview
  • Context-free Commands
    • Compare
    • Convert
    • Detect
    • Info
    • Download
    • Explain
    • Explore
    • Filter
    • Generate
    • Merge
    • Patch
    • Prune
    • Stats
    • Transform
    • Validate
  • Context Commands
    • Checkout
    • Commit
    • Create
    • Export
    • Log
    • Model (context)
    • Project (context)
    • Source (context)
    • Status
    • Util
  • Helper Commands
    • Format

API Reference

  • Datumaro Module
    • datumaro
      • datumaro.cli
        • datumaro.cli.commands
        • datumaro.cli.contexts
        • datumaro.cli.helpers
        • datumaro.cli.util
      • datumaro.components
        • datumaro.components.abstracts
        • datumaro.components.algorithms
        • datumaro.components.annotation
        • datumaro.components.annotations
        • datumaro.components.cli_plugin
        • datumaro.components.comparator
        • datumaro.components.config
        • datumaro.components.config_model
        • datumaro.components.contexts
        • datumaro.components.crypter
        • datumaro.components.dataset
        • datumaro.components.dataset_base
        • datumaro.components.dataset_item_storage
        • datumaro.components.dataset_storage
        • datumaro.components.environment
        • datumaro.components.errors
        • datumaro.components.exporter
        • datumaro.components.extractor_tfds
        • datumaro.components.filter
        • datumaro.components.format_detection
        • datumaro.components.generator
        • datumaro.components.hl_ops
        • datumaro.components.importer
        • datumaro.components.launcher
        • datumaro.components.lazy_plugin
        • datumaro.components.media
        • datumaro.components.merge
        • datumaro.components.operations
        • datumaro.components.progress_reporting
        • datumaro.components.project
        • datumaro.components.registry
        • datumaro.components.shift_analyzer
        • datumaro.components.transformer
        • datumaro.components.validator
        • datumaro.components.visualizer
      • datumaro.errors
      • datumaro.ops
      • datumaro.plugins
        • datumaro.plugins.accuracy_checker_plugin
        • datumaro.plugins.anchor_generator
        • datumaro.plugins.configurable_validator
        • datumaro.plugins.data_formats
        • datumaro.plugins.explorer
        • datumaro.plugins.framework_converter
        • datumaro.plugins.inference_server_plugin
        • datumaro.plugins.missing_annotation_detection
        • datumaro.plugins.ndr
        • datumaro.plugins.openvino_plugin
        • datumaro.plugins.sam_transforms
        • datumaro.plugins.sampler
        • datumaro.plugins.specs
        • datumaro.plugins.splitter
        • datumaro.plugins.synthetic_data
        • datumaro.plugins.tiling
        • datumaro.plugins.transforms
        • datumaro.plugins.validators
      • datumaro.project
      • datumaro.rust_api
      • datumaro.util
        • datumaro.util.annotation_util
        • datumaro.util.attrs_util
        • datumaro.util.definitions
        • datumaro.util.file_utils
        • datumaro.util.image
        • datumaro.util.image_cache
        • datumaro.util.import_util
        • datumaro.util.log_utils
        • datumaro.util.mask_tools
        • datumaro.util.meta_file_util
        • datumaro.util.multi_procs_util
        • datumaro.util.os_util
        • datumaro.util.pickle_util
        • datumaro.util.points_util
        • datumaro.util.samples
        • datumaro.util.scope
        • datumaro.util.tabular_util
        • datumaro.util.telemetry_stub
        • datumaro.util.telemetry_utils
        • datumaro.util.tf_util
      • datumaro.version
  • Supported Plugins

Explanation

  • Concepts
  • Architecture
  • Command-Line Workflow
  • Plugins
    • OpenVINO™ Inference Interpreter

Misc

  • How to use Datumaro
  • Model Preparation
  • Extending
  • How to control telemetry data collection

Release Notes

  • Release Notes
  • Docs
  • From...

From Datumaro to Model Training#

Here we provide E2E examples from Datumaro to model trainers.

Train Your OpenVINO™ Model Using YoloV8 Trainer For Any Dataset Format

Noisy label detection for classification tasks from training loss dynamics statistics

Noisy label detection for detection tasks from training loss dynamics statistics

Missing Annotation Detection

previous

Automatic instance mask annotations generation using Segment Anything Model

next

Train Your OpenVINO™ Model Using YoloV8 Trainer For Any Dataset Format

Show Source

Created using Sphinx 7.2.6.

Built with the PyData Sphinx Theme 0.15.2.