CV-Studio is a graphical annotation tool to address different Computer Vision tasks.

CV-Studio is developed in Python, Qt, SQLite and uses PyTorch's resources to train deep learning models.

CVStudio supports:

Datasets:

• Create and manage your datasets for images.
• Manually annotate images:
  • Using a label system for classification problems.
  • Using a bounding box for localization and object detection problems.
  • Using a polygon tool or freehand selection for segmentation tasks.
• Auto-annotate images with a pretrained model to continue tagging the images by your own.

Watch a demo video

• Datasets: Annotations for videos.
• Models:
  • Build your own custom models using a pretrained model from PyTorch Hub and your annotated dataset.
  • Publish your own custom models to PyTorch Hub.
• Experiments: Develop experiments using your datasets and models from PyTorch Hub or your custom trained models.
• Platforms: macOS and Linux support.

Note: CV-Studio only have been developed and tested on Windows. Future platforms (macOS and Linux) are in the roadmap.

Windows + Anaconda:

• Clone the repository:

    git clone https://github.com/haruiz/CvStudio.git

• Download and install Anaconda (Python 3+).

• Open Anaconda Prompt, go to CvStudio directory and follow the next steps:

  • Create a new environment with Python 3.6:

    conda create --name cvstudio python=3.6

  • Install required libraries:

    pip install matplotlib
    pip install numpy
    pip install opencv-contrib-python
    pip install pillow
    pip install tqdm
    pip install scipy
    pip install "dask[complete]"
    pip install more-itertools
    pip install pandas
    pip install PyQt5
    pip install imutils
    pip install peewee
    pip install -U marshmallow
    pip install hurry.filesize
    pip install Mako

  • Install PyTorch with conda following the instructions in the official site. For the purpose of this tutorial we are selecting the following configuration:

    • Using GPU:
      

        conda install pytorch torchvision cudatoolkit=10.0 -c pytorch-nightly

    • Using CPU:
      

        conda install pytorch torchvision cpuonly -c pytorch-nightly -c defaults -c conda-forge

This command must be executed from the CVStudio folder: Windows (PowerShell)

   Invoke-WebRequest -OutFile ./models/MS_DeepLab_resnet_trained_VOC.pth https://data.vision.ee.ethz.ch/csergi/share/DEXTR/MS_DeepLab_resnet_trained_VOC.pth
   Invoke-WebRequest -OutFile ./models/dextr_pascal-sbd.pth https://data.vision.ee.ethz.ch/csergi/share/DEXTR/dextr_pascal-sbd.pth

Linux

    wget https://data.vision.ee.ethz.ch/csergi/share/DEXTR/MS_DeepLab_resnet_trained_VOC.pth -P ./models
    wget https://data.vision.ee.ethz.ch/csergi/share/DEXTR/dextr_pascal-sbd.pth -P ./models

    python cvstudio.py

Check out the wiki.

Send a pull request.

Free software: MIT license

Citation: haruiz. CV-Studio. Git code (2019). https://github.com/haruiz/CvStudio

• Icons made by Dave Gandy from www.flaticon.com
• Icons made by Pixelmeetup from www.flaticon.com
• Icons made by Smashicons from www.flaticon.com
• Icons made by Eucalyp from www.flaticon.com
• Icons made by Becris from www.flaticon.com
• Icons made by Smashicons from www.flaticon.com
• Icons made by Freepik from www.flaticon.com
• Icons made by Those Icons from www.flaticon.com

Deep Extreme Cut: From Extreme Points to Object Segmentation