lazylabel-gui 1.3.6

An image segmentation GUI for generating ML ready mask tensors and annotations.

LazyLabel

AI-Assisted Image Segmentation for Machine Learning Applications

LazyLabel integrates Meta's Segment Anything Model (SAM) with advanced editing capabilities to enable efficient, high-precision image annotation. Designed for computer vision research and machine learning dataset preparation.

---

Quick Start

Installation

pip install lazylabel-gui
lazylabel-gui

Optional: SAM 2.1 Support

For enhanced performance with SAM 2.1 models:

pip install git+https://github.com/facebookresearch/sam2.git

Note: SAM 2.1 is optional - LazyLabel downloads SAM 1.0 models by default.

Usage

1. Open Folder - Select your image directory
2. AI Segmentation - Click on objects for automatic mask generation
3. Manual Refinement - Edit polygons, merge segments, adjust classifications
4. Export - Generate .npz files with one-hot encoded masks for ML training

---

Key Features

• One-click AI segmentation with Meta's SAM and SAM 2.1 models
• Manual polygon drawing with full vertex control
• Smart editing tools - merge segments, adjust class names, and class order
• ML-ready exports - One-hot encoded .npz format and .json for YOLO format
• Image enhancement - brightness, contrast, gamma adjustment
• Advanced image viewer - zoom, pan, and real-time adjustments
• Edge cropping - define custom crop areas to focus on specific regions
• Undo/Redo system - full history of all actions
• Auto-saving - automatic saving of labels when navigating between images
• Advanced filtering - FFT thresholding and color channel thresholding
• Customizable hotkeys for all functions

---

Essential Hotkeys

Action	Key	Description
AI Mode	1	Point-click segmentation
Draw Mode	2	Manual polygon drawing
Edit Mode	E	Select and modify shapes
Save Segment	Space	Confirm current mask
Merge	M	Combine selected segments
Pan	Q + drag	Navigate large images
Positive Point	Left Click	Add to segment
Negative Point	Right Click	Remove from segment

Note: All hotkeys are fully customizable - Click "Hotkeys" button to personalize your workflow.

---

Output Format

LazyLabel exports data in standardized formats optimized for machine learning workflows:

import numpy as np

# Load your labeled data
data = np.load('your_image.npz')
mask = data['mask']  # Shape: (height, width, num_classes)

# Each channel is a binary mask for one class
class_0_mask = mask[:, :, 0]  # Background
class_1_mask = mask[:, :, 1]  # Object type 1
class_2_mask = mask[:, :, 2]  # Object type 2

Ideal for:

• Semantic segmentation datasets
• Instance segmentation training
• Computer vision research
• Automated annotation pipelines

---

Development

Requirements: Python 3.10+, ~2.5GB disk space for SAM models (auto-downloaded)

Installation from Source

git clone https://github.com/dnzckn/LazyLabel.git
cd LazyLabel
pip install -e .
lazylabel-gui

Testing & Quality

# Run test suite (272 tests)
pytest --tb=short

# Code quality checks
ruff check --fix src/

Architecture

• Modular design with clean component separation
• Signal-based communication between UI elements
• Extensible model system for SAM 1.0 and SAM 2.1 variants
• Comprehensive test suite (272 tests with extensive coverage)
• Multi-view support for simultaneous image processing

---

Contributing

LazyLabel welcomes contributions! Please review:

• Usage Manual for comprehensive user documentation
• Architecture Guide for technical implementation details
• Issues page for feature requests and bug reports

---

Acknowledgments

• LabelMe
• Segment-Anything-UI