# Tao mine od images

> Tao mine od images is an open-source agent skill for design and media. Mines unique source images closest to target samples for object detection datasets. It is available on Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, Windsurf and Goose.

- **Kind:** Skill
- **Pricing:** Open source
- **Platforms:** Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, Windsurf, Goose
- **Website:** https://github.com/NVIDIA/skills/tree/main/skills/tao-mine-od-images
- **Repository:** https://github.com/nvidia/skills
- **GitHub stars:** 4,037
- **License:** Apache-2.0
- **Skill name:** tao-mine-od-images
- **Collection:** https://github.com/NVIDIA/skills
- **Categories:** design and media, developer tools
- **Listing updated:** 2026-11-18

## About

This skill runs TAO Data Services unique-neighbor matching mining, which selects a set of distinct source images whose embeddings sit nearest to target samples. It works from precomputed embedding parquet files; the skill does not generate embeddings, so upstream steps must produce them with the same encoder for both sides.

The workflow centers on a YAML spec that names the source and target parquets, an output directory and the desired count of unique files. The skill includes a verification script that checks the spec before a Docker run, a template with default tuning values, and a launch command that mounts a shared run root so host and container paths match. Outputs include a parquet of selected filepaths and a summary.json with coverage statistics.

Class-stratified allocation requires a rare class list plus source and target detection files in COCO JSON or KITTI format, and a CUDA GPU with Docker and the NVIDIA container toolkit is needed to run the miner.

## When your agent uses it

> Run TAO Data Services TMM unique-neighbor matching mining from embedding parquet files for object detection workflows. Use when an object detection workflow needs to mine a bijectively-assigned set of unique source images closest to target samples. Use global allocation when mining without class constraints. Use class_stratified when rare classes are specified.

## What it can access

From a static check of the files on GitHub; nothing was run.

- [Runs Python scripts](https://github.com/NVIDIA/skills/tree/HEAD/skills/tao-mine-od-images)
- [Asks for: Read, Bash](https://github.com/NVIDIA/skills/blob/HEAD/skills/tao-mine-od-images/SKILL.md)
- [No network calls found in its scripts](https://github.com/NVIDIA/skills/tree/HEAD/skills/tao-mine-od-images)
- [License: Apache-2.0](https://github.com/NVIDIA/skills/tree/HEAD/skills/tao-mine-od-images)

## Features

- **Global unique matching:** Selects source images closest to targets without class constraints, using a bijective assignment.
- **Class-stratified allocation:** Balances the selection toward specified rare classes using COCO or KITTI detection labels.
- **Spec verification:** A script validates the YAML spec and reports budget, policy and metric before the run starts.
- **Spec template:** Provides a default configuration file with tuning values for the mining run.
- **Run summary output:** Writes coverage percentage and per-class breakdowns to a summary JSON file.
- **Docker launch workflow:** Gives the commands to pull the pinned data-services image and run it with GPU access.

## Good for

- Pick a unique set of training images closest to a target distribution for object detection
- Mine extra images that cover rare classes for an object detection dataset
- Exclude already-labeled images from a new mining run
- Check how much of the requested unique count a mining run actually reached

## Install

### Claude Code

Copy the skill folder to `~/.claude/skills/tao-mine-od-images` to use it in all your projects:

```sh
tmp=$(mktemp -d) &&
git clone --depth 1 https://github.com/NVIDIA/skills.git "$tmp" &&
mkdir -p ~/.claude/skills &&
rm -rf ~/.claude/skills/tao-mine-od-images &&
cp -R "$tmp/skills/tao-mine-od-images" ~/.claude/skills/tao-mine-od-images &&
rm -rf "$tmp"
```

For one project only, put the folder in `.claude/skills/tao-mine-od-images` inside it instead.

### Codex

Copy the skill folder to `~/.agents/skills/tao-mine-od-images` to use it in every repository:

```sh
tmp=$(mktemp -d) &&
git clone --depth 1 https://github.com/NVIDIA/skills.git "$tmp" &&
mkdir -p ~/.agents/skills &&
rm -rf ~/.agents/skills/tao-mine-od-images &&
cp -R "$tmp/skills/tao-mine-od-images" ~/.agents/skills/tao-mine-od-images &&
rm -rf "$tmp"
```

For one repository only, put the folder in `.agents/skills/tao-mine-od-images` inside it instead.

### Cursor

Copy the skill folder to `~/.cursor/skills/tao-mine-od-images` to use it in every project:

```sh
tmp=$(mktemp -d) &&
git clone --depth 1 https://github.com/NVIDIA/skills.git "$tmp" &&
mkdir -p ~/.cursor/skills &&
rm -rf ~/.cursor/skills/tao-mine-od-images &&
cp -R "$tmp/skills/tao-mine-od-images" ~/.cursor/skills/tao-mine-od-images &&
rm -rf "$tmp"
```

For one project only, put the folder in `.cursor/skills/tao-mine-od-images` inside it instead.

### Gemini CLI

Run in a terminal (add `--scope workspace` to install it for the current workspace only):

```sh
gemini skills install https://github.com/NVIDIA/skills.git --path skills/tao-mine-od-images
```

Or copy the folder to `~/.gemini/skills/tao-mine-od-images` yourself.

### VS Code

For GitHub Copilot in VS Code, copy the skill folder to `~/.copilot/skills/tao-mine-od-images`:

```sh
tmp=$(mktemp -d) &&
git clone --depth 1 https://github.com/NVIDIA/skills.git "$tmp" &&
mkdir -p ~/.copilot/skills &&
rm -rf ~/.copilot/skills/tao-mine-od-images &&
cp -R "$tmp/skills/tao-mine-od-images" ~/.copilot/skills/tao-mine-od-images &&
rm -rf "$tmp"
```

For one project only, put the folder in `.github/skills/tao-mine-od-images` inside it instead.

### Other agents

Get the folder:

```sh
git clone --depth 1 https://github.com/NVIDIA/skills.git
```

Then copy `skills/skills/tao-mine-od-images` into your agent's skills folder as `tao-mine-od-images`. Agents that support SKILL.md read one folder per skill.

## FAQ

### What does this skill need before it can run?

Precomputed source and target embedding parquet files from the same encoder, Docker, the NVIDIA container toolkit and at least one CUDA GPU.

### Does it compute embeddings?

No. Upstream steps must produce the source and target embedding parquets before this skill runs.

### What does class_stratified mode require?

A rare class list, source and target detection files, and a detection format of coco or kitti; the spec verifier rejects the policy without them.

### Where are the results written?

A final_unique_files.parquet with selected filepaths and a summary.json with coverage stats, in the configured output directory.

### How do I install Tao mine od images in Claude Code?

Copy the skill folder from GitHub to ~/.claude/skills/tao-mine-od-images to use it in all your projects, or to .claude/skills/tao-mine-od-images in one project.

### Does Tao mine od images run scripts?

Yes. It runs Python scripts. No network calls were found in its scripts. This comes from a static check of the files on GitHub; nothing was run. Read the files before you install it.

### Which AI agents can use Tao mine od images?

Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, Windsurf and Goose. Each reads skills in the SKILL.md format from its own folder; the install steps on this page list them.

### What are the best alternatives to Tao mine od images?

The closest alternatives to Tao mine od images are Skill creator, Reverse engineer anything, Rea changelog update, Rea ci tests audit and Rea analysis optimization.

## Where this comes from

- Description: from the skill's own SKILL.md.

## Alternatives

- [Skill creator](https://ai-powered.com/skills/skill-creator): Skill creator is a free agent skill for ai models and llm tooling.
- [Reverse engineer anything](https://ai-powered.com/skills/reverse-engineer-anything): Reverse engineer anything is an open-source agent skill for security.
- [Rea changelog update](https://ai-powered.com/skills/rea-changelog-update): Rea changelog update is an open-source agent skill for developer tools.
- [Rea ci tests audit](https://ai-powered.com/skills/rea-ci-tests-audit): Rea ci tests audit is an open-source agent skill for developer tools.
- [Rea analysis optimization](https://ai-powered.com/skills/rea-analysis-optimization): Rea analysis optimization is an open-source agent skill for developer tools.
- [Rea tool design](https://ai-powered.com/skills/rea-tool-design): Rea tool design is an open-source agent skill for design and media.
- [Improve codebase architecture](https://ai-powered.com/skills/improve-codebase-architecture): Improve codebase architecture is an open-source agent skill for developer tools.
- [Prototype](https://ai-powered.com/skills/prototype): Prototype is an open-source agent skill for developer tools.

---
Source: https://ai-powered.com/skills/tao-mine-od-images
