segment-image（分割图像）

Segment Image

Segment and identify objects in images using Gemini's vision capabilities.

When to Use

Use this skill when the user asks to:

• Identify objects in an image
• Generate masks for specific objects
• Segment an image into regions
• Extract objects from an image

Usage

bun run --cwd ${CLAUDE_PLUGIN_ROOT} ${CLAUDE_PLUGIN_ROOT}/skills/segment-image/scripts/segment.ts <input-image> [options]

Options

• --prompt <text> - Custom segmentation prompt
• --output <dir> - Output directory for mask files

Examples

# Segment all objects
bun run --cwd ${CLAUDE_PLUGIN_ROOT} ${CLAUDE_PLUGIN_ROOT}/skills/segment-image/scripts/segment.ts photo.jpg

# Segment with custom prompt
bun run --cwd ${CLAUDE_PLUGIN_ROOT} ${CLAUDE_PLUGIN_ROOT}/skills/segment-image/scripts/segment.ts photo.jpg --prompt "identify all people and vehicles"

# Save masks to directory
bun run --cwd ${CLAUDE_PLUGIN_ROOT} ${CLAUDE_PLUGIN_ROOT}/skills/segment-image/scripts/segment.ts photo.jpg --output ./masks

Context Discipline

Do not read generated mask images back into context. The script outputs file paths. Ask the user to visually inspect the masks. To inspect programmatically, optimize the images first (via the optimize-images skill).

Model

Uses gemini-3-flash-preview (Gemini 3 Flash) for image segmentation.

Last verified: February 2026. If a newer generation exists, STOP and suggest a PR to b-open-io/gemskills. See the ask-gemini skill's references/gemini-api.md for current models and Google's official gemini-api-dev skill for the canonical source.