A Python library for every Computer Vision Engineer
Project description
CvPal ๐ผ๏ธ
๐ Introduction
Welcome to CvPal - Your Ultimate Computer Vision Companion! ๐
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.
๐ Proudly Sponsored by VisionFullSpace ๐
๐ฏ Features
- ๐ผ๏ธ Synthetic Image Generation: Create custom datasets using text prompts
- ๐ Dataset Merging: Seamlessly combine datasets with multiple attributes
- ๐๏ธ Label Management: Effortlessly remove or replace specific labels from your dataset
- ๐ Label Occurrence Counting: Accurately track label frequencies
- ๐ Dataset Reporting: Generate comprehensive reports on your dataset
- ๐ Easy Integration: Smoothly integrate with existing ML pipelines
- ๐ Comprehensive Documentation: Detailed guides for all features
- ๐ค AI-Powered Dataset Creation: Generate entire datasets with just a prompt
- ๐ผ Pandas Integration: Group images and labels into a pandas DataFrame for advanced manipulation
๐ Dataset Folder Structure
For optimal performance, please structure your dataset folder as follows:
folder/
โโโ train/
โ โโโ images/
โ โโโ labels/
โโโ test/
โ โโโ images/
โ โโโ labels/
โโโ valid/
โโโ images/
โโโ labels/
For TXT format datasets, include a data.yaml
config file:
folder/
โโโ 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
๐ Installation
Install CvPal with a simple pip command:
pip install cvpal==1.0.2
๐ง 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()
๐ Supported Models and Formats
Model Name | Supported Format | Support in Package | Detection | Segmentation |
---|---|---|---|---|
YOLOv5-v11 | TXT & YAML config | โ | โ | โ |
YOLOv3-v4 | Darknet TXT | โ | โ | โ |
EfficientDet | Pytorch JSON annotations | โ | โ | โ |
Detectron 2 | JSON annotations | โ | โ | โ |
Segmentation Models | XML format | โ | โ | โ |
TensorFlow Object Detection | Binary format | โ | โ | โ |
Fine-tune PaliGemma | JSONL annotations | โ | โ | โ |
Apple's CreateML | Proprietary JSON format | โ | โ | โ |
Turi Create tools | Proprietary JSON format | โ | โ | โ |
๐ค Call for Contributions
Join the CvPal community and make a global impact! We welcome contributions of all sizes:
- ๐ Bug fixes and enhancements
- ๐ Documentation improvements
- ๐จ UI/UX enhancements
- ๐งช New feature development
To contribute major changes, please reach out through our mailing list first.
Other ways to contribute:
- ๐ Help triage issues
- ๐ Create tutorials and presentations
- ๐ต๏ธ Review pull requests
๐ License
This project is licensed under the MIT License - see the LICENSE file for details.
Empowering Computer Vision Worldwide
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
File details
Details for the file cvpal-1.0.2.tar.gz
.
File metadata
- Download URL: cvpal-1.0.2.tar.gz
- Upload date:
- Size: 15.7 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/5.1.0 CPython/3.11.9
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 83ec2df6b71ae06aca51c35deb61aff230f76188081c743a689f906f04e81f58 |
|
MD5 | 1fcab366815420c1ceda0e0943cc5a2d |
|
BLAKE2b-256 | c190024964b7b30dd9aed173da9c987d5ced2a8abd12c2452caa15d3afe3a527 |
File details
Details for the file cvpal-1.0.2-py3-none-any.whl
.
File metadata
- Download URL: cvpal-1.0.2-py3-none-any.whl
- Upload date:
- Size: 16.2 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/5.1.0 CPython/3.11.9
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | a57d124359a45b583ccd8a5d6a6c47271b760e0db925724764e0e097632545c6 |
|
MD5 | 7de0bd78ab5f6c88071c38115a6ff25d |
|
BLAKE2b-256 | 3b350c849ff8970f9b2eb11732b721c82c6b20424fe3c448ed390213420d15e6 |