cvpal 1.0.2

A Python library for every Computer Vision Engineer

CvPal 🖼️

🌟 Introduction

Welcome to CvPal - Your Ultimate Computer Vision Companion! 🚀

CvPal is a cutting-edge Python package designed to empower Computer Vision engineers worldwide. Our mission is to streamline image operations, dataset management, and synthetic data generation, allowing you to focus on what truly matters: building and optimizing your machine learning pipelines.

🏆 Proudly Sponsored by VisionFullSpace 🏆

🎯 Features

• 🖼️ Synthetic Image Generation: Create custom datasets using text prompts
• 🔄 Dataset Merging: Seamlessly combine datasets with multiple attributes
• 🗑️ Label Management: Effortlessly remove or replace specific labels from your dataset
• 📊 Label Occurrence Counting: Accurately track label frequencies
• 📝 Dataset Reporting: Generate comprehensive reports on your dataset
• 🔌 Easy Integration: Smoothly integrate with existing ML pipelines
• 📚 Comprehensive Documentation: Detailed guides for all features
• 🤖 AI-Powered Dataset Creation: Generate entire datasets with just a prompt
• 🐼 Pandas Integration: Group images and labels into a pandas DataFrame for advanced manipulation

📁 Dataset Folder Structure

For optimal performance, please structure your dataset folder as follows:

folder/
├── train/
│   ├── images/
│   └── labels/
├── test/
│   ├── images/
│   └── labels/
└── valid/
    ├── images/
    └── labels/

For TXT format datasets, include a data.yaml config file:

folder/
└── data.yaml

Example data.yaml:

names:
  - Old_Paper
  - Rock
  - Scissors
nc: 3
roboflow:
  license: Private
  project: rock-paper-scissors-sxsw
  url: https://universe.roboflow.com/roboflow-58fyf/rock-paper-scissors-sxsw/dataset/14
  version: 14
  workspace: roboflow-58fyf
test: ../test/images
train: Rock-Paper-Scissors-SXSW-14/train/images
val: Rock-Paper-Scissors-SXSW-14/valid/images

🚀 Installation

Install CvPal with a simple pip command:

pip install cvpal==1.0.2

🔧 Example Usage

from cvpal.generate import DetectionDataset
from cvpal.preprocessing import ImagesDetection

# Generate synthetic data
detection_dataset = DetectionDataset(model="stable-diffusion")
detection_dataset.generate("a cat looking at the camera", num_images=2, labels=["cat"], output_type="yolo", overwrite=False)

# Preprocess existing dataset
cp = ImagesDetection()
cp.read_data("/content/Rock-Paper-Scissors-SXSW-14", data_type="txt")
cp.report()

📊 Supported Models and Formats

Model Name	Supported Format	Support in Package	Detection	Segmentation
YOLOv5-v11	TXT & YAML config	✅	✅	✅
YOLOv3-v4	Darknet TXT	❌	❌	❌
EfficientDet	Pytorch JSON annotations	❌	❌	❌
Detectron 2	JSON annotations	❌	❌	❌
Segmentation Models	XML format	❌	❌	❌
TensorFlow Object Detection	Binary format	❌	❌	❌
Fine-tune PaliGemma	JSONL annotations	❌	❌	❌
Apple's CreateML	Proprietary JSON format	❌	❌	❌
Turi Create tools	Proprietary JSON format	❌	❌	❌

🤝 Call for Contributions

Join the CvPal community and make a global impact! We welcome contributions of all sizes:

• 🐛 Bug fixes and enhancements
• 📝 Documentation improvements
• 🎨 UI/UX enhancements
• 🧪 New feature development

To contribute major changes, please reach out through our mailing list first.

Other ways to contribute:

• 🔍 Help triage issues
• 📚 Create tutorials and presentations
• 🕵️ Review pull requests

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

---

Powered by VisionFullSpace
Empowering Computer Vision Worldwide