# Roboflow integration

> Computer vision platform API: manage workspaces, projects, datasets, training and Workflows, and run hosted model and Workflow inference.

- Authentication: apiKey

## Tools (177)

- **Action Recognition**: Classify the actions in a video clip.
- **Add Projects To Folder**: Adds projects to a folder.
- **Agent: Chat with the agent**: Send a message to the agent.
- **Agent: Get a conversation**: Return the full conversation including all messages.
- **Agent: List conversations**: Return all agent conversations in the workspace.
- **Agent: Publish a Workflow**: Deploy the latest draft version of a Workflow that the agent created or edited.
- **Attach Inference Metadata**: Attaches custom metadata to inference results (by `inference_id` from an inference response) for the Model Monitoring dashboard.
- **Batch Create Vision Events**: Create up to 100 vision events in a single request.
- **Batch Update Image Metadata**: Updates metadata and tags on up to 1,000 images as an async task (202).
- **Cancel Training**: Cancels a queued or running job.
- **Cancel Training Run**: Cancels a queued v2 training run.
- **CLIP Compare**: Run the Open AI CLIP model to compute similarity scores.
- **CLIP Image Embeddings**: Run the Open AI CLIP model to embed image data.
- **CLIP Text Embeddings**: Run the Open AI CLIP model to embed text data.
- **Compare Model Evaluations**: Compare models on one Dataset Version.
- **Create Annotation Job**: Assigns images from a batch to a labeler and reviewer.
- **Create API Key**: Creates a key.
- **Create Auto Label Job**: Starts a paid auto-labeling job over an image batch.
- **Create a Vision Event**: Create a single vision event.
- **Create Project**: Creates a project.
- **Create Project Folder**: Creates a project folder.
- **Create Training**: Starts a training run on a version through the v2 trainings API; consumes credits.
- **Create Workflow**: Creates a workflow.
- **Dedicated: Create Deployment**: Host `https://roboflow.cloud` (not api.roboflow.com).
- **Dedicated: Delete Deployment**: Host `https://roboflow.cloud` (not api.roboflow.com).
- **Dedicated: Get Deployment**: Host `https://roboflow.cloud` (not api.roboflow.com).
- **Dedicated: Get Deployment Usage**: Host `https://roboflow.cloud` (not api.roboflow.com).
- **Dedicated: Get Logs**: Host `https://roboflow.cloud` (not api.roboflow.com).
- **Dedicated: Get Workspace Usage**: Host `https://roboflow.cloud` (not api.roboflow.com).
- **Dedicated: List Deployments**: Host `https://roboflow.cloud` (not api.roboflow.com).
- **Dedicated: List Machine Types**: Host `https://roboflow.cloud` (not api.roboflow.com).
- **Dedicated: Pause Deployment**: Host `https://roboflow.cloud` (not api.roboflow.com).
- **Dedicated: Resume Deployment**: Host `https://roboflow.cloud` (not api.roboflow.com).
- **Dedicated: Update Startup Script**: Host `https://roboflow.cloud` (not api.roboflow.com).
- **Delete Dataset Version**: Moves one dataset version to Trash (kept 30 days).
- **Delete Images**: Removes images from a project.
- **Delete Project**: Moves a project to Trash (kept 30 days) and cancels its in-flight training jobs.
- **Delete Project Folder**: Deletes a folder; its projects return to the top level and are not deleted.
- **Delete Training**: Moves a non-running training and its models to Trash (kept 30 days).
- **Delete Vision Events**: Deletes up to 1,000 events from a use case.
- **Delete Workflow**: Moves a workflow to Trash (kept 30 days).
- **Depth Estimation**: Run the depth estimation model to generate a depth map.
- **Depth Estimation (Model In Path)**: Run depth estimation.
- **Device Manager: Create a Device**: Create a new device in the workspace and return the identifiers needed to install it.
- **Device Manager: Get a Device**: Retrieve a single device by ID.
- **Device Manager: Get a Stream**: Retrieve a single stream on a device.
- **Device Manager: Get Default Device Config**: Returns the workspace default configuration applied to new Deployment Manager devices.
- **Device Manager: Get Deployment Activity**: Returns the device's activity history.
- **Device Manager: Get Device Config**: Returns the device's current runtime configuration.
- **Device Manager: Get Device Events**: Returns device and stream lifecycle events (for example device boots, stream starts and stops, errors, config changes).
- **Device Manager: Get Device Logs**: Returns paginated device logs.
- **Device Manager: Get Device Service**: Returns one service, in the same shape as an entry in the list.
- **Device Manager: Get Device Service Metric Series**: Returns one named metric for one service over time, for charting.
- **Device Manager: Get Device Snapshot**: Returns the current state of a device in one call: the device record, streams, services, latest telemetry, Docker and NTP health, hardware, issues from the current deployment, and recent events.
- **Device Manager: Get Device Telemetry**: Returns device health data: hardware metrics (CPU, memory, disk, GPU, network) bucketed across the requested time window, plus Docker and NTP health, per-service container status, per-stream FPS.
- **Device Manager: Get Exception Occurrences**: Returns the raw occurrences behind one exception fingerprint from the deployments response, newest first.
- **Device Manager: Get Pending Commands**: Returns the commands queued for a device, oldest first.
- **Device Manager: List Config Revisions**: List prior configuration revisions for a device, newest first.
- **Device Manager: List Devices**: List every device registered in the workspace.
- **Device Manager: List Device Services**: Returns the services a device runs and what each one is doing now.
- **Device Manager: List Streams**: List every stream configured on a device, including streams that are set up in the device config but have not reported in yet, which are returned with a provisioning status.
- **Device Manager: Queue a Command**: Queue a command for a device.
- **Device Manager: Report Pipeline Status**: Report the inference pipelines a device is running.
- **Disable Active Learning**: Turns Active Learning off.
- **DocTR OCR**: Run the DocTR OCR model to retrieve text in an image.
- **EasyOCR OCR**: Run the EasyOCR model to retrieve text in an image.
- **Enable Active Learning**: Turns Active Learning on; selects a model if none is set.
- **Export Dataset Version**: Returns a download link for the version exported in a format (e.g.
- **Fork Universe Project**: Forks a public Universe project into the workspace.
- **Fork Workflow**: Copies a workflow from another workspace into this one.
- **Generate Dataset Version**: Creates a new dataset version (frozen snapshot with preprocessing and augmentation).
- **Generate Workflow Token**: Generates a short-lived token for running a workflow from a public client such as a browser.
- **Get Active Learning Config**: Returns the Active Learning configuration and deployment state.
- **Get Annotation Insights**: Returns per-labeler annotation metrics from 2023-08-01 onward.
- **Get Annotation Insights (Legacy)**: Legacy annotation insights endpoint, slated for deprecation; prefer Get Annotation Insights.
- **Get Annotation Job**: Returns one annotation job.
- **Get API Key**: Returns masked metadata for one key.
- **Get Async Task Status**: Polls any async task (`taskId` + `url` returned by 202 endpoints).
- **Get Auto Label Job Status**: Returns progress of an auto-label job.
- **Get a Vision Event Video Playback URL**: Return a URL that plays one prepared video, valid for 60 minutes.
- **Get Billing Usage Report**: Returns credit usage by API key prefix and feature.
- **Get Dataset Health**: Returns dataset quality statistics.
- **Get Dataset Version**: Returns one dataset version without listing the project's other versions.
- **Get Dataset ZIP Upload Status**: Polls a zip upload task.
- **Get Evaluation Confidence Sweep**: Returns precision/recall/F1 per confidence threshold and the F1-optimal threshold per split and class.
- **Get Evaluation Confusion Matrix**: Returns the confusion matrix; `matrix[actual][predicted]` is an instance count (pixel count for segmentation).
- **Get Evaluation Image Predictions**: Returns paginated per-image TP/FP/FN stats, cluster id and 2D embedding.
- **Get Evaluation mAP Results**: Returns per-split primary metrics.
- **Get Evaluation Performance By Class**: Returns per-class metrics for one split.
- **Get Evaluation Recommendations**: Returns improvement recommendations.
- **Get Evaluation Vector Analysis**: Returns UMAP + HDBSCAN clustering of image embeddings with per-cluster metrics.
- **Get Image Batch**: Returns one batch.
- **Get Image Details**: Returns details of one image or video, including its annotation boxes and URLs.
- **Get Image Search Export Status**: Returns the status of an image search export task.
- **Get Inference Stats**: Returns Model Monitoring statistics for deployed models, with the previous period for comparison.
- **Get Model Evaluation**: Returns one evaluation; done ones include headline metrics.
- **Get Plan Info**: Returns the workspace plan information.
- **Get Project**: Returns one project with its metadata, class counts and all versions.
- **Get Project Base Workflow**: Returns the project's base Workflow, selected model and Active Learning state.
- **Get Publishable Key**: Returns the workspace publishable key (`rf_<workspaceId>`).
- **Get Training**: Returns one training run on a version, with its status, models and metrics.
- **Get Training Recipe**: Returns the accepted training recipe for a version.
- **Get Training Results**: Returns the training summary for a version; NAS runs include child models, hardware recommendations and mining data.
- **Get Video Upload Status**: Reads the processing status of a video upload.
- **Get Vision Event Video Status**: Read the processing status of a video prepared for Vision Events.
- **Get Workflow**: Returns a workflow's specification and metadata.
- **Get Workflow Evals OpenAPI**: Returns the workspace's complete live OpenAPI document for the Workflow Evals API (Specs, Datasets, Cases, Runs, exports, embeddings), with request schemas, scopes and headers.
- **Get Workspace**: Returns workspace metadata and every project in it.
- **Grounding DINO Detect**: Run the Grounding DINO zero-shot object detection model.
- **List Active Learning Images**: Lists the Active Learning review queue.
- **List Annotation Jobs**: Lists the annotation jobs in a project.
- **List API Keys**: Lists the workspace API keys (masked) and the publishable key.
- **List Auto Label Models**: Lists foundation models usable for Auto Label.
- **List Image Batches**: Lists the upload batches in a project.
- **List Model Evaluations**: Lists model evaluations in a workspace.
- **List Project Folders**: Lists project folders.
- **List Project Models**: Lists every trained model in a project, including version models and standalone (NAS child) models.
- **List Trainings**: Lists the training runs on a dataset version.
- **List Trash**: Lists items in Trash with their cleanup dates.
- **List Workflow Eval Evaluators**: Lists evaluator types and config schemas of the installed Workflow Evals engine.
- **List Workflows**: Lists workflows in the workspace.
- **List Workflow Versions**: Returns the versioned snapshots of a workflow specification.
- **OWLv2 Detect**: Run the google owlv2 model to few-shot object detect.
- **Perception Encoder Compare**: Run the Meta Perception Encoder model to compute similarity scores.
- **Perception Encoder Image Embeddings**: Run the Meta Perception Encoder model to embed image data.
- **Perception Encoder Text Embeddings**: Run the Meta Perception Encoder model to embed text data.
- **PP-OCR**: Run PP-OCRv6 two-stage OCR to retrieve text in an image.
- **Prepare a Vision Event Video Upload**: Create a pending video and return a signed URL to PUT the file to.
- **Prepare Video Upload**: Creates a pending video and returns a signed URL for its file.
- **Preview Auto Label**: Runs one image through a foundation model without creating a job.
- **Query Vision Events**: Query vision events with filters, time ranges, and pagination.
- **Rebalance Splits**: Rebalances train/valid/test splits on the project; the next generated version uses them.
- **Remove Projects From Folder**: Removes projects from a folder back to the workspace top level.
- **Restore From Trash**: Restores a project, version, training or workflow.
- **Revoke API Key**: Permanently revokes a key; apps using it fail immediately.
- **Run Model Inference**: Runs a trained model (object detection, instance segmentation, classification, keypoints, semantic segmentation) on an image.
- **Run Multimodal Model**: Run inference with the specified large multi-modal model.
- **Run Multimodal Model (Model In Path)**: Run inference with the specified large multi-modal model.
- **Run Workflow**: Runs a saved Workflow on `https://serverless.roboflow.com`.
- **SAM2 Image Embeddings**: Run the Meta AI Segment Anything 2 Model to embed image data.
- **SAM2 Image Segmentation**: Run the Meta AI Segment Anything 2 Model to generate segmenations for image data.
- **SAM3 3D Object Generation**: Generate 3D meshes and Gaussian splatting from 2D images with mask prompts.
- **SAM3 Concept Segmentation**: Run the SAM3 PCS (promptable concept segmentation) to generate segmentations for image data.
- **SAM3 Image Embeddings**: Run the Model to embed image data.
- **SAM3 Visual Segmentation**: Run the SAM3 PVS (promptable visual segmentation) to generate segmentations for image data.
- **SAM Image Embeddings**: Run the Meta AI Segmant Anything Model to embed image data.
- **SAM Image Segmentation**: Run the Meta AI Segmant Anything Model to generate segmenations for image data.
- **Search Project Images**: Searches images in one project by semantic similarity, tag, class, batch or job.
- **Search Universe**: Searches public Roboflow Universe projects.
- **Search Universe (POST)**: Searches public Roboflow Universe projects with a JSON body.
- **Search Workspace Images**: Searches images across the whole workspace with the same query syntax as the app (e.g.
- **Select Project Base Workflow Model**: Sets the model used by the project's base Workflow.
- **Start Dataset ZIP Upload**: Starts an async dataset upload (zip up to 2 GB and 10,000 files, images plus annotations in COCO, YOLO, VOC and other formats).
- **Start Image Search Export**: Starts an async export of workspace image search results.
- **Start Training Job**: Starts a hosted training run on a version; consumes credits.
- **Stop Training Early**: Requests an early stop of a running job; weights so far are kept.
- **Stop Training Run**: Stops an active v2 training run.
- **TrOCR OCR**: Run the TrOCR model to retrieve text in an image.
- **Update Active Learning Config**: Sets collection limits and filters.
- **Update API Key**: Updates a key; send only the fields to change.
- **Update Image Metadata**: Updates metadata and tags on one image synchronously.
- **Update Image Tags**: Adds, removes, or replaces the tags on one image.
- **Update Project Folder**: Updates a folder.
- **Update Workflow**: Replaces a workflow's name, URL and specification.
- **Upload Annotation**: Attaches an annotation file to an uploaded image, in any supported format that references the image file name.
- **Upload a Vision Event Bundle**: Create one vision event from a self-contained tarball bundle written by the Write Vision Event Bundle workflow block.
- **Upload a Vision Event Image**: Upload an image for use in vision events.
- **Upload Image**: Uploads one image to a project, as multipart `file`, a base64 request body, or a hosted `image` URL.
- **Verify Credentials**: Confirms the API key works and names the workspace it belongs to.
- **Vision Events: Archive a Use Case**: Archive a use case by setting its status to inactive.
- **Vision Events: Create a Use Case**: Create a new use case in your workspace.
- **Vision Events: Get Custom Metadata Schema**: Retrieve the schema of custom metadata fields for a given use case.
- **Vision Events: Get Image Metadata Schema**: Retrieve the per-image metadata keys used by events in a given use case.
- **Vision Events: List Use Cases**: List all use cases that have recorded vision events in your workspace.
- **Vision Events: Unarchive a Use Case**: Restore a previously archived use case by setting its status back to active.
- **Vision Events: Update a Use Case**: Update the name or status of an existing use case.
- **YOLO-World Detect**: Run the YOLO-World zero-shot object detection model.

Connect Roboflow in General Input: https://www.generalinput.com/apps/roboflow