← 返回 Skills 市场

Watermark Remover

Name: Watermark Remover
Author: fusae

作者 fusae · GitHub ↗ · v0.1.0 · MIT-0

cross-platform ⚠ suspicious

总下载

当前安装

版本数

在 OpenClaw 中安装

/install fusae-watermark-remover

功能描述

Automatically detects and removes watermarks from images using AI-powered inpainting. Use when user asks to "remove watermark", "clean image", or "去水印".

使用说明 (SKILL.md)

Watermark Remover

Automatically detects watermarks in image corners and removes them using LaMa AI model.

Installation

This skill requires the watermark-remover Python package. Install it first:

# Install from PyPI (when published)
pip install watermark-remover

# Or install from source
git clone https://github.com/yourusername/watermark-remover.git
cd watermark-remover
pip install -e .

Dependencies: Python 3.9+, opencv-python, numpy, Pillow, iopaint

Script Directory

Scripts in scripts/ subdirectory. Replace ${SKILL_DIR} with this SKILL.md's directory path.

Script	Purpose
`scripts/main.py`	Watermark removal CLI wrapper

Usage

python ${SKILL_DIR}/scripts/main.py \x3Cinput> [options]

Options

Option	Description	Default
`\x3Cinput>`	Image file or directory	Required
`--output`	Output path	`\x3Cinput>/no_watermark`
`--corner-ratio`	Corner area ratio (0-1)	0.15
`--threshold`	Detection sensitivity (lower = more sensitive)	30
`--padding`	Mask dilation pixels	10
`--preview`	Generate mask preview only	false
`--no-lama`	Use OpenCV instead of LaMa	false

Examples

# Single image
python ${SKILL_DIR}/scripts/main.py image.jpg

# Directory with custom output
python ${SKILL_DIR}/scripts/main.py ./photos/ --output ./cleaned/

# Preview detection (shows red marks on watermarks)
python ${SKILL_DIR}/scripts/main.py ./photos/ --preview

# Adjust detection sensitivity
python ${SKILL_DIR}/scripts/main.py image.jpg --corner-ratio 0.2 --threshold 20

# Use OpenCV inpaint (faster but lower quality)
python ${SKILL_DIR}/scripts/main.py image.jpg --no-lama

How It Works

Scans image corners (default 15% area)
Detects watermarks using edge detection + high-frequency analysis
Generates precise mask
Removes watermark using LaMa AI model (auto-downloads on first run)
Falls back to OpenCV inpaint if LaMa fails

Supported Formats

JPG, JPEG, PNG, BMP, TIFF, WEBP

安全使用建议

This skill appears to do what it says, but it depends on an external 'watermark-remover' Python package that is not provided or verified in the registry (the docs use a placeholder GitHub URL). Installing that package (from PyPI or a git repo) can run arbitrary code and will auto-download large ML model files at runtime. Before installing or using: (1) verify the package source — prefer an official GitHub repo or PyPI page and inspect its code; (2) inspect the 'iopaint' and 'watermark-remover' packages for malicious behaviors or unexpected network access; (3) run installation and execution in an isolated environment (VM/container) if you must test it; (4) if you cannot verify the upstream project, do not install system-wide; (5) remember removing watermarks can violate copyright/terms — ensure you have the right to modify images. If you provide the official package URL or the upstream repo, a re-evaluation could raise confidence.

能力评估

ℹ Purpose & Capability

The name/description match the behavior: the skill detects/removes watermarks and uses LaMa/inpainting. Requiring opencv/numpy/Pillow/iopaint is plausible. However, the skill's code is only a thin wrapper that invokes an external 'watermark-remover' CLI rather than implementing in-place functionality, and the SKILL.md points to a placeholder GitHub repo (https://github.com/yourusername/...) and no homepage/source is provided in registry metadata — this is an unexpected gap between the shipped artifact and its external dependency.

ℹ Instruction Scope

Runtime instructions and scripts only check for an installed 'watermark-remover' binary and then run it. They do not request unrelated files or environment variables. Caveat: running the external CLI and its auto-download of the LaMa model means the skill will perform network activity and execute code outside the skill's bundle at runtime.

⚠ Install Mechanism

No install spec is declared in the registry even though the skill requires installing an external Python package (pip install watermark-remover or git clone). Installing a PyPI package or code from an arbitrary GitHub repo can execute arbitrary code on the host. The repo URL in docs is a placeholder (yourusername) and there's no verified source/homepage, which increases supply-chain risk. The listed dependency 'iopaint' is uncommon and should be vetted before installation.

✓ Credentials

The skill requests no environment variables, credentials, or config paths. That is proportionate to the stated purpose.

✓ Persistence & Privilege

always is false and no special privileges are requested. The wrapper will execute an installed CLI on invocation; autonomous invocation is allowed by default but not combined here with broad credential or persistence requests.

如何使用

确保已安装 OpenClaw（本地或 Docker 部署）
在对话框中输入安装命令：/install fusae-watermark-remover
安装完成后，直接呼叫该 Skill 的名称或使用 /fusae-watermark-remover 触发
根据 Skill 的参数说明提供必要输入，即可获得结构化输出

版本历史

v0.1.0

Initial release of Watermark Remover skill. - Automatically detects and removes watermarks in image corners using AI-powered inpainting (LaMa), with fallback to OpenCV. - Command-line interface provided via `scripts/main.py`, supporting multiple options for detection sensitivity, masking, and output locations. - Includes a preview mode to visualize detected watermarks. - Supports common image formats: JPG, PNG, BMP, TIFF, and WEBP. - Recommended for use when asked to “remove watermark”, “clean image”, or “去水印”.

元数据

Slug fusae-watermark-remover

版本 0.1.0

许可证 MIT-0

累计安装 0

当前安装数 0

历史版本数 1

常见问题