Four-step image referring-expression pipeline: turns images plus KITTI bounding-box labels into region descriptions, scene captions, grounded referring expressions, and (optionally) verified expressions via VLM distillation.
Install
npx skills add NVIDIA/skills --skill tao-generate-referring-expressionsnpx skills add NVIDIA/skills --skill tao-generate-referring-expressions -a claude-code -gnpx skills add NVIDIA/skills --skill tao-generate-referring-expressions -a codex -ggemini skills install https://github.com/NVIDIA/skills.git --path skills/tao-generate-referring-expressionsgh skill install NVIDIA/skills tao-generate-referring-expressionsgit clone --depth 1 https://github.com/NVIDIA/skills /tmp/NVIDIA__skills
mkdir -p ~/.claude/skills ~/.agents/skills
cp -R /tmp/NVIDIA__skills/skills/tao-generate-referring-expressions ~/.claude/skills/tao-generate-referring-expressions # Claude Code
cp -R /tmp/NVIDIA__skills/skills/tao-generate-referring-expressions ~/.agents/skills/tao-generate-referring-expressions # Codex, Cursor, Gemini CLI, Copilot, OpenCode