🌟 TrainCV 🌟

No-code Labeling and Training Toolkit for Computer Vision

With Improved Labelme for Image Labeling

TODO

This project is under development. Please consider everything here unstable. There are a lot of features need to be added in the future.

You can request new features through this contact form.

• Labeling: Integrate labelme
• Labeling: UI for textbox labeling (OCR, labels + positions)
• Labeling: Group objects (can be used in key-value matching problems)
• Labeling: Auto-labeling with YOLOv5
• Labeling: Tracking for video labeling
• Training: Project + Experiment management
• Training: Object detection
• Training: Image classification
• Training: Image segmentation
• Training: Instance segmentation
• Training: Add docker support for training
• Deployment: Export to ONNX
• Deployment: Export to TFLite
• Deployment: Export to TensorRT
• CI/CD for Pypi package publishment
• Unit tests
• Documentation

I. Install and run

• Requirements: Python >= 3.8

• Recommended: Miniconda/Anaconda https://docs.conda.io/en/latest/miniconda.html

• Create environment:

conda create -n traincv python=3.8
conda activate traincv

• (For macOS only) Install PyQt5 using Conda:

conda install -c conda-forge pyqt==5.15.7

• Install traincv:

pip install traincv

• Run app:

traincv_app

Or

python -m traincv.app

II. Development

• Generate resources:

pyrcc5 -o traincv/resources/resources.py traincv/resources/resources.qrc

• Run app:

python traincv/app.py

III. References

• labelme
• gpu_util
• Icons: Flat Icons