wkentaro/labelme

Image annotation with Python. Supports polygon, rectangle, circle, line, point, and AI-assisted annotation.

What it solves

Labelme 是一个圖形化圖像標註工具，旨在幫助用戶建立用於電腦視覺任務的 ground-truth 數據。簡化了為各種 AI 模型（例如用於對象物體檢測和分割的模型）標註圖像的過程。

How it works

該工具使用 Python 編寫並使用 Qt framework 進行界面，允許用戶在圖像上手動繪製形狀（多邊形、矩形、圓形、線條和點）以定義對象。它還集成了 SAM (Segment Anything Model)、EfficientSAM 和 YOLO-world 以實現更快的 point-to-polygon/mask 和 text-to-annotation 工作流。標註信息 lautsav_repo_name_placeholder_here_saved as JSON 檔案，隨後可導出至 VOC 和 COCO 等常見數據集格式。

Who it's for

它適用於需要為圖像分類、邊界框檢測、語義分割和實例分割建立高品質標註數據集的研究人員和開發人員。

Highlights

  • Diverse Annotation Primitives: Supports polygon, rectangle, circle, line, and point tools.

  • AI-Assisted Labeling: Integrates SAM, package-name-placeholder-here_integrates SAM, EfficientSAM, and YOLO-world for automated mask and text-based annotation.

  • Multi-format Export: Exports data to VOC and COCO formats for semantic and instance segmentation.

  • Video Annotation: Includes support for annotating video frames.

  • Divergent Accessibility: Available in

  • Standalone App: Offers a standalone executable for an users who do not want to manage Python or Qt dependencies.

    hought: 1. Analyze the Request: The user wants a translation of a project summary and body into four languages: Traditional Chinese (zh-TW), Simplified Chinese (zh-CN), Japanese (ja), and Korean (ko). The output must be a single, valid, parseable JSON object following a specific schema. No extra text, no comments, no control tokens (like `

or ` inside the strings, though the prompt says