Roboflow Data Management

For agents — source-of-truth: This skill is authored in `roboflow/computer-vision-skills` and shipped with the Roboflow plugin. If your client has loaded the plugin (you'll see roboflow:<name> skills in your available skills list), use those local skills — they're read fresh from disk every session. The same content served as MCP resources at roboflow://skills/<name>/... is a fallback for clients without the plugin and may lag this repo. Don't call `ReadMcpResourceTool` for `roboflow://skills/...` URIs when a local `roboflow:<name>` skill is available.

Data Management on Roboflow

Project Types

Project type is set at creation and cannot be changed later.

Uploading Data

Methods

Limits: Max 20 MB per image, max 16,400 x 10,900 px. Duplicate images are skipped automatically.

Video Upload

Videos are split into frames at a configurable rate (1 frame/60s to 60 fps). Supported formats depend on browser (MP4 H.264 most compatible).

CLI Upload

pip install roboflow
roboflow import -w <workspace> -p <project-id> /path/to/dataset

Tags

Tags are free-form labels on images for organization and filtering.

Dataset Search (RoboQL)

Search images via the Images page search bar. Combine filters with boolean logic.

Filters

Boolean Logic

• AND, OR, NOT, parentheses: class:helmet AND NOT (tag:v1 OR tag:v2)
• Inverted filter with -: -class:vest
• Comparison operators on numeric filters: >, <, >=, <=, = (e.g., class:helmet>=3)

Splits (Train / Valid / Test)

Images are assigned to train, valid, or test splits. Splits are rebalanced during version generation (Step 2 in version creation). Augmentations only apply to train split.

Dataset Versions

A version is a frozen snapshot of the dataset at a point in time. Changes to the project after version creation do not affect existing versions.

Version Creation Pipeline

1. Source selection — images from the dataset split 2. Train/Test split — rebalance percentages 3. Preprocessing — applied to all splits (train + valid + test) 4. Augmentation — applied only to train split 5. Generate — creates immutable version

Preprocessing Options

Augmentation Options

Applied to train images only. Configurable max version size (e.g., 3x = source + 2 augmented copies).

Dataset Analytics

Available at project sidebar -> "Analytics". Shows:

• Image count, annotation count, avg image size, median aspect ratio
• Missing and null annotation counts
• Class distribution across train/valid/test
• Image dimension insights (size + aspect ratio distribution)
• Annotation heatmap (click-drag to filter images by region)
• Object count histogram (click bars to see matching images)

Classes

Managed at Project Settings -> Classes.

Warning: Class changes at project level affect all images (irreversible). Use version-level "Modify Classes" preprocessing for non-destructive changes.

Annotation Groups

Annotation group = the category encompassing all classes in a project. Projects sharing the same annotation group share their class list and annotations.

• Enable during project creation: "Share image annotations with other projects"
• Shared annotations: editing in one project affects all linked projects
• Look for chain-link icon to identify shared images/projects
• Images shared across projects count only once toward usage

Project Folders

Folders group projects for organization. SSO workspaces can restrict folder access to specific team members.

Export Formats

Versions can be exported as .zip download or curl command. 40+ formats supported including COCO, YOLO, Pascal VOC, TFRecord, and more. Full list at roboflow.com/formats.

Export via Python SDK:

project.version(1).download("yolov8")

MCP apps vs plain tools

Prefab MCP apps (create_project_app) exist when parameters are unclear, you need real UX, or a human must confirm after seeing form fields — plain chat/MCP calls should not guess project type and license alone.

MCP Tools Available

Related Pages

• roboflow://skills/roboflow-labeling/SKILL — annotation tools, AI labeling, Label Assist, Smart Polygon, Auto Label, annotation jobs

Labeling & Annotation on Roboflow

Source-of-truth note: This page ships with the Roboflow plugin. If your client has the plugin loaded, prefer the local skill (roboflow:data-management) over fetching roboflow://skills/data-management/labeling via ReadMcpResourceTool — the MCP resources are a fallback for non-plugin clients and may lag the source repo.

Annotation Tools

Annotation Type Compatibility

_*Polygons/masks auto-converted to bounding boxes for object detection._

Multi-Select & Context Menu

Hold Shift + click or drag-select multiple annotations. Right-click for bulk actions:

Other Controls

• Undo / Redo — standard undo/redo while in B, P, or S mode
• Repeat Previous — reapply last annotations in same positions
• Zoom — bottom-left zoom tool, supports zoom lock
• Class Selector — appears on annotation; type to filter/create classes

AI Labeling Features

All AI labeling features consume credits. See roboflow.com/credits for rates.

Label Assist Details

1. Open image in annotator -> click magic wand 2. Select model: "Your Models" tab or "Public Models" tab (star on Universe first) 3. Configure class mapping (remap model classes to project classes) 4. Predictions appear as you navigate images

Smart Polygon (SAM) Details

1. Enable Smart Polygon (S) -> select "Enhanced" 2. Hover to preview mask, click to create 3. Refine: click outside mask to expand, inside to shrink 4. Toggle polygon complexity: Convex Hull / Smooth / Complex 5. Press Enter to accept

Box Prompting Details

1. Draw at least one bbox example per class 2. Activate Box Prompting tool 3. Predictions appear as dotted lines; adjust confidence slider 4. Right-click false positives -> "Convert to Negative" 5. Click "Approve Predictions" to save 6. Model improves as you annotate more images in the session

Limitation: Best with images under 1000px; degrades on 2000px+ with small objects.

Auto Label Details

1. Upload images -> select "Auto Label" 2. Define classes + optional text descriptions 3. "Generate Test Results" on 4-image subset (free, no credits) 4. Adjust confidence per class 5. "Auto Label with This Model" to run on full batch

Models used: Grounding DINO (detection), Grounded SAM (segmentation), CLIP (classification), or your own Roboflow-trained model.

Annotation Workflow

Upload -> Batch (Unassigned) -> Assign Job -> Annotating -> Review -> Dataset
                                                  ^                    |
                                                  |--- Rejected -------|

Batches

Each upload creates a batch. Batches track images through the pipeline.

Batches can only be deleted from Unassigned. Move back first if needed.

Jobs & Assignment

• Assign a batch (or subset) to one or more team members
• Unassigned images in a partially-assigned batch split into a new batch
• Add labeling instructions before assigning ("Add Instructions" -> "Edit")
• Labelers receive notifications when assigned

Review Mode

Reviewer sees images in Approved / Rejected / To Do tabs:

• Approve — image moves to Dataset
• Reject — image returns to Annotating for rework

Collaboration Features

Annotation Groups

Projects sharing the same annotation group share classes and annotations. Useful for:

• Multiple detection tasks on the same images (e.g., "chess pieces" vs "board games")
• Shared images counted once toward usage
• Editing annotations in one project affects all linked projects

Enable at project creation: "Share image annotations with other projects". Look for the chain-link icon on shared images/projects.

MCP Tools Available