data preparation for GUI automation

Hey Product Hunt! 👋 Screph is a desktop tool for capturing and annotating screenshots with purpose—not just to share, but to use them: for documentation, QA, training data, or feeding context to LLMs. Why Screph? Most screenshot tools stop at "draw a rectangle and add an arrow." But if you've ever tried to document a complex UI, prepare assets for AI coding assistants, or create consistent training datasets, you know it's not enough. What makes it different: 📐 Smart Selection: Choose how you select elements — from simple color picker to advanced segmentation (GrabCut, Watershed, neural nets). Pick the right level of precision for your task. 🗣️ Voice Descriptions: Annotate hands-free. Speak, and your words become structured metadata attached to selected regions. 🧩 UI Hierarchy: Capture not just pixels, but relationships between elements (parent/child). Great for documenting nested components. 📤 Structured Export: Output clean JSON with coordinates, labels, and hierarchy — ready to feed into code generators, test frameworks, or LLMs. 🔒 Offline & Private: Speech recognition (Vosk) and CV models run locally. Your data stays on your machine. 100% Open Source (AGPLv3). Built with Python & Qt. Would love to hear how you'd use it!