Roboflow APIroboflow.com ↗
Search and retrieve metadata from 200,000+ open-source computer vision datasets on Roboflow Universe. Access class labels, image counts, and version info.
What is the Roboflow API?
This API provides access to Roboflow Universe's catalog of over 200,000 open-source computer vision datasets across 3 endpoints. Use search_datasets to query by keyword and get paginated results with image counts, class labels, and workspace slugs, or use get_dataset_details to pull the full metadata for any specific dataset — including class distribution, version history, and download links.
curl -X GET 'https://api.parse.bot/scraper/4d4cf0a1-b912-46f8-9ca4-54eb70b69488/search_datasets?page=1&query=dogs' \ -H 'X-API-Key: $PARSE_API_KEY'
Typed, relational, agent-ready
A generated client with real types, enums, and the links between objects — the structure a flat JSON response can't carry. Autocompletes in your editor and reads cleanly to coding agents.
- Fully typed · autocompletes
- Objects link to objects
- Typed errors & pagination
Typed Python client. Set up the SDK in your uv project, then pull this API’s typed client:
uv add parse-sdk uv run parse init uv run parse add --marketplace roboflow-com-api
uv run parse add --marketplace pulls a pinned snapshot of this canonical API — it won’t change underneath you. To customize it, subscribe and swap to your own copy.
"""Walkthrough: Roboflow Universe SDK — search, browse, and inspect CV datasets."""
from parse_apis.Roboflow_Universe_API import RoboflowUniverse, DatasetNotFound
client = RoboflowUniverse()
# Search for dog-related datasets, capped at 5 results
for summary in client.dataset_summaries.search(query="dogs", limit=5):
print(summary.name, summary.image_count, summary.classes)
# Drill into the first result's full details via the instance method
summary = client.dataset_summaries.search(query="chess", limit=1).first()
if summary:
detail = summary.details()
print(detail.name, detail.image_count, detail.description[:100])
# Fetch a dataset directly by workspace + project slug
try:
dataset = client.datasets.get(workspace="brad-dwyer", project="chess-pieces-new")
print(dataset.name, dataset.latest_version, dataset.classes)
except DatasetNotFound as exc:
print(f"Dataset not found: {exc.workspace}/{exc.project}")
# Browse car-related datasets
for car_ds in client.dataset_summaries.list_cars(page_limit=1, limit=3):
print(car_ds.name, car_ds.image_count, car_ds.workspace_url)
print("exercised: dataset_summaries.search / summary.details / datasets.get / dataset_summaries.list_cars")
Full-text keyword search across all Roboflow Universe datasets. Returns up to 50 dataset summaries per page. Results include workspace and project slugs needed to fetch full details via get_dataset_details. Paginates via integer page counter; has_more indicates whether additional pages exist.
| Param | Type | Description |
|---|---|---|
| page | integer | Page number for pagination (1-based). |
| query | string | Search keyword. Supports single words and multi-word phrases. |
{
"type": "object",
"fields": {
"page": "integer, current page number",
"datasets": "array of dataset summaries each containing id, name, workspace, workspace_url, project_slug, image_count, classes, description_short, download_page_url, and dataset_url",
"has_more": "boolean, true if more pages are available"
},
"sample": {
"data": {
"page": 1,
"datasets": [
{
"id": "xsuEOP8QAweeccNwwObI",
"name": "dogs",
"classes": [
"Welsh Corgi"
],
"workspace": {
"id": "J1p0ztU8K5WlJg34umfCNrw7qsL2",
"url": "tracy-h8ob2",
"name": "Tracy"
},
"dataset_url": "https://universe.roboflow.com/tracy-h8ob2/dogs-gzdcf",
"image_count": 109,
"project_slug": "dogs-gzdcf",
"workspace_url": "tracy-h8ob2",
"description_short": "dogs",
"download_page_url": "https://universe.roboflow.com/tracy-h8ob2/dogs-gzdcf/dataset/1/download"
}
],
"has_more": true
},
"status": "success"
}
}About the Roboflow API
Search and Browse Datasets
The search_datasets endpoint accepts a query string (single words or multi-word phrases) and an optional page integer for pagination. Each result in the returned datasets array includes id, name, workspace, project_slug, image_count, classes, description_short, and download_page_url. The has_more boolean tells you whether additional pages exist. Results are capped at 50 datasets per page.
Bulk Retrieval with list_car_datasets
The list_car_datasets endpoint automates multi-page fetching for car-related datasets in a single call. Pass page_limit to control how many pages are retrieved (up to 50 datasets per page each). The response returns a flat datasets array with the same structure as search_datasets results, plus a total_fetched count — useful when you need a larger working set without manually looping through pages.
Dataset Detail Retrieval
The get_dataset_details endpoint requires a workspace slug and a project slug, both of which are returned by search_datasets under workspace.url and project_slug respectively. The response exposes the full description in markdown, classes as a keyed object with annotation counts per class, latest_version, image_count, dataset_url, and download_page_url. This is the endpoint to call when you need to evaluate whether a specific dataset fits your labeling requirements before downloading.
The Roboflow API is a managed, monitored endpoint for roboflow.com — not a raw scraper you maintain. Every endpoint is automatically health-checked on a schedule, and when roboflow.com changes and a check fails, the API is automatically queued for repair and re-verified. It is built to keep working as the site underneath it changes.
This isn't an official roboflow.com API — it's an independent, maintained REST wrapper over public data. Where the source has no official API (or only a limited one), Parse gives you a stable contract over a source that never promised one, and keeps it current. Need a new endpoint or field? You can revise it yourself in plain English and the agent rebuilds it against the live site in minutes — contributing the change back to the shared API is free.
Will this API break when the source site changes?+
Is this an official API from the source site?+
Can I fix or extend this API myself if I need a new endpoint or field?+
What happens if I call an endpoint that has an issue?+
- Find labeled datasets for a specific object class by querying
search_datasetswith a keyword and filtering byclasses. - Compare annotation volume across similar datasets using
image_countand class annotation counts fromget_dataset_details. - Build a dataset discovery tool that surfaces
description_shortandimage_countfor CV practitioners browsing by domain. - Automate dataset auditing by retrieving
latest_versionandclassesdistributions for a set of project slugs. - Aggregate a list of car detection datasets with annotation counts using
list_car_datasetsbefore benchmarking models. - Generate download manifests by collecting
download_page_urlfields across multiple search result pages. - Identify active workspaces publishing datasets in a niche domain using the
workspacefield in search results.
| Tier | Price | Credits/month | Rate limit |
|---|---|---|---|
| Free | $0/mo | 100 | 5 req/min |
| Hobby | $30/mo | 1,000 | 20 req/min |
| Developer | $100/mo | 5,000 | 100 req/min |
One credit = one API call regardless of which marketplace API you call. Exceeding the rate limit returns a 429 response. Authenticate with the X-API-Key header.
Does Roboflow have an official developer API?+
What does get_dataset_details return beyond what search_datasets already provides?+
get_dataset_details returns the full markdown description, per-class annotation counts (as an object with class names as keys), latest_version, and a direct download_page_url. The search_datasets endpoint only provides description_short and a flat classes list, making get_dataset_details necessary when you need annotation distribution data or want to evaluate dataset quality before use.Is there a limit to how many datasets search_datasets returns per request?+
has_more boolean in the response indicates whether additional pages exist, and you increment the page parameter to fetch them. If you need larger batches without manual pagination, list_car_datasets accepts a page_limit parameter that fetches multiple pages in one call.Does the API expose dataset download files or model weights?+
download_page_url and dataset_url — that points to where files can be accessed, but does not serve dataset files or model artifacts directly. You can fork this API on Parse and revise it to add an endpoint that resolves download links or retrieves version-specific export formats.Can I search datasets by class label or filter by image count?+
search_datasets endpoint accepts a keyword query but does not expose class-label or image-count filters as dedicated parameters. You can retrieve a result set and filter on the client side using the classes and image_count fields returned in each dataset summary. For more targeted filtering, you can fork this API on Parse and revise it to add parameter-level filtering logic.