Support Center
Frequently Asked Questions
Everything you need to know about the platform, pricing, and more.
Getting Started
cvPal is a computer vision dataset platform that gives developers and teams powerful tools to curate, process, and manage datasets — all through a managed MCP Server. The platform includes 40+ dataset operations, multi-format exports (YOLO, COCO, Pascal VOC, HuggingFace, and more), team collaboration, and integrations with Hugging Face, Kaggle, and Roboflow.
Sign up for a free account at cvpal.dev, verify your email, and you'll have immediate access to the free tier. From your dashboard you can create datasets, run operations, and start making MCP tool calls. Our documentation includes step-by-step guides and tutorials to help you get productive quickly.
cvPal supports 14 dataset types including object detection, image classification, semantic segmentation, instance segmentation, keypoint estimation, OCR, video tracking (MOT), 3D point clouds, vision-language, and tabular data. Each type has a dedicated viewer and type-aware operations.
Yes. The free tier includes 10 MCP Server requests per month and 500 MB of storage — enough for learning, prototyping, and small projects. You can upgrade to a paid plan at any time to unlock higher limits and additional features.
Click 'New Dataset' in your dashboard, choose the dataset type, and upload your files. You can also import directly from Roboflow, Kaggle, or Hugging Face. Once created, use the 40+ built-in operations to clean, augment, split, merge, and export your data.
cvPal provides a Python SDK, REST APIs, and a managed MCP Server that works with Claude, Cursor, Windsurf, Gemini, OpenAI, and any MCP-compatible AI agent. You can also import datasets from Hugging Face, Kaggle, and Roboflow, and export to Google Drive or Google Cloud Storage.