A high-performance utility for extracting specific objects from image datasets based on standard bounding box annotation formats.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install image-cropper
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install image-cropper using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Image Cropper?
Image Cropper is a versatile tool designed for developers and data scientists working with computer vision datasets. It automates the process of isolating objects from larger images using existing annotation files in COCO, YOLO, VOC, or LabelMe formats. By integrating this into your Openclaw Skills collection, you can rapidly prepare training data or extract visual assets with precision.
This skill streamlines the workflow of dataset preparation, allowing for batch processing and custom padding to ensure that extracted objects retain necessary context for further analysis or model training. It is an essential utility for anyone needing to transform raw annotated data into clean, object-centric image assets.
Image Cropper Use Cases
- Extracting individual objects from complex scenes for training image classification models.
- Generating localized image assets from large-scale computer vision datasets for web or app display.
- Preparing cropped image datasets for quality assurance of bounding box annotations.
- Creating sprite sheets or galleries of specific detected objects from diverse source imagery.
How Image Cropper Works
- The user provides a source directory containing images and corresponding annotation files in a supported format like YOLO, COCO, VOC, or LabelMe.
- The skill parses the metadata to identify the precise coordinates of all bounding boxes within each image.
- For each identified bounding box, the script applies optional padding and minimum size filters to ensure high-quality extraction.
- The isolated regions are cropped using the Pillow library and saved to the specified output directory as individual image files.
Image Cropper Setup
To get started with this component of Openclaw Skills, you must first install the required image processing library:
pip install pillow
Once installed, you can execute crops using the following command structure:
# Example for YOLO format
python scripts/cropper.py yolo images/ labels/ output/
# Example for COCO format
python scripts/cropper.py coco annotations.json images/ output/
Image Cropper Data Schema & Taxonomy
The skill handles input and output data across several standard AI formats to maintain compatibility with major labeling tools:
| Component | Description |
|---|---|
| Input Formats | COCO (JSON), YOLO (TXT), VOC (XML), LabelMe (JSON) |
| Image Inputs | Standard image formats including JPG and PNG |
| Output Files | Individual object crops named by source image and object index |
| Metadata | Preserves object relationships and allows for batch processing tracking |
Image Cropper Advanced Features
- Custom Padding: Add specific pixel margins around extracted objects to provide necessary visual context for downstream tasks.
- Size Filtering: Define a minimum box size to skip low-resolution or insignificant detections automatically.
- Format Conversion: Force output crops into specific formats like JPEG or PNG regardless of the source file type.
- Batch Processing: Efficiently handle thousands of images in a single command execution for large-scale Openclaw Skills implementations.
- Quality Control: Adjust JPEG quality settings (1-100) to balance storage space and image fidelity.
SKILL.md
Loading
Related Openclaw Skills
Image Annotation QC
An automated quality control tool for validating, scoring, and visualizing image annotation accuracy in computer vision datasets.
Dataset Splitter
A robust utility for automatically partitioning image datasets and annotations into training, validation, and testing subsets.
Daily Horoscope & Almanac
A specialized agent for querying traditional Chinese Almanac data, providing daily auspicious and inauspicious guidance.
Daily Fengshui Attire
A metaphysical dressing consultant that calculates the luckiest clothing colors based on the Chinese Earthly Branches and Five Elements of the day.
Image Deduplicator
A high-performance tool for detecting and managing duplicate or visually similar images using perceptual hashing.
Image Quality Filter
An automated computer vision tool designed to detect and filter out blurry, poorly lit, or low-resolution images from datasets.
Featured*
