# Pictograph

> Developer-first computer vision platform. Annotate images with SAM3, train custom models, and deploy inference APIs from one API, SDK, and CLI.

Pictograph turns directories of images into trained, deployed computer-vision models with minimal hand-annotation. The same REST API backs a typed Python SDK ("pip install pictograph"), a CLI, and a 37-tool agent registry for Claude and OpenAI. Pricing is USD-denominated compute credits with a free tier and no credit card required.

## Product

- [Auto-annotation (SAM3)](https://pictograph.io/auto-annotate): Point, box, and text prompts; single image or async batch over a dataset. Outputs editable polygons and boxes.
- [Training](https://pictograph.io/train): Train YOLOX, RF-DETR, Segformer, and classification models on managed GPUs. Returns model weights.
- [Deployments](https://pictograph.io/deployments): Serve a trained model as one authenticated /predict endpoint, billed by uptime, with pause and resume.
- [Workflows](https://pictograph.io/workflows): Visual node-graph pipelines over images, video, or datasets with object tracking, counting, dwell time, and webhook sinks.
- [Explore](https://pictograph.io/explore): Public datasets and trained models to star, comment on, and fork.
- [Pricing](https://pictograph.io/pricing): USD compute credits, four tiers, free to start.
- [Glossary](https://pictograph.io/glossary): Plain-language definitions of computer-vision terms - annotation shapes, dataset formats, training, accuracy metrics, and inference.

## Documentation

- [Docs home](https://pictograph.io/docs)
- [Quickstart](https://pictograph.io/docs/quick-start)
- [Installation](https://pictograph.io/docs/installation)
- [API reference](https://pictograph.io/docs/api-reference)
- [Agents (Claude and OpenAI)](https://pictograph.io/docs/agents)
- [Docs index for LLMs](https://pictograph.io/docs/llms.txt)
- [Full documentation bundle](https://pictograph.io/docs/llms-full.txt)

## Guides

- [Blog](https://pictograph.io/blog): Practical guides on annotation, training, and shipping vision models.
- [Polygon vs bounding box annotation](https://pictograph.io/blog/polygon-vs-bounding-box): When to use each, with tradeoffs.
- [Train YOLOX and export to ONNX](https://pictograph.io/blog/train-yolox-export-onnx): A no-infrastructure walkthrough.

## Datasets and models

- [Explore](https://pictograph.io/explore): Browse public datasets and trained models to star, comment on, and fork.
- [Public datasets](https://pictograph.io/datasets): Curated, downloadable computer-vision datasets (COCO, YOLO, Pascal VOC, CVAT, and more).
- [Trained models](https://pictograph.io/models): Public models with ONNX weights and reported metrics.
- [Browse by task](https://pictograph.io/explore/task): Public datasets and models grouped by CV task.
- [Object detection datasets and models](https://pictograph.io/explore/task/object-detection): Bounding-box datasets and YOLOX / RF-DETR detectors.
- [Image segmentation datasets and models](https://pictograph.io/explore/task/segmentation): Polygon-mask datasets and instance / semantic segmentation models.
- [Keypoint and pose datasets and models](https://pictograph.io/explore/task/keypoint-detection): COCO-keypoint datasets and RF-DETR keypoint models.
- [Pictograph on Hugging Face](https://huggingface.co/pictograph): Datasets and models mirrored to the Hub with ready-to-run weights.

## Compare

- [Pictograph vs Roboflow](https://pictograph.io/compare/roboflow-alternative)
- [Pictograph vs CVAT](https://pictograph.io/compare/cvat-alternative)
- [Pictograph vs V7](https://pictograph.io/compare/v7-alternative)
- [Pictograph vs Labelbox](https://pictograph.io/compare/labelbox-alternative)
- [Pictograph vs Ultralytics](https://pictograph.io/compare/ultralytics-hub-alternative)
- [Pictograph vs SuperAnnotate](https://pictograph.io/compare/superannotate-alternative)
- [Pictograph vs Encord](https://pictograph.io/compare/encord-alternative)
- [Pictograph vs Supervisely](https://pictograph.io/compare/supervisely-alternative)
- [Switching from Roboflow](https://pictograph.io/compare/switching-from-roboflow): step-by-step dataset migration guide.
- [Switching from V7](https://pictograph.io/compare/switching-from-v7): step-by-step V7 Darwin dataset migration guide.

## Optional

- [GitHub](https://github.com/pictograph-io)
- [Python SDK on GitHub](https://github.com/pictograph-io/pictograph-sdk)
- [Hugging Face](https://huggingface.co/pictograph)
- [PyPI](https://pypi.org/project/pictograph/)
- [Agent tool registry (JSON Schema)](https://api.pictograph.io/api/v1/developer/tools.json)