- Git
- Ubuntu-based OS (Windows is not supported at this time. Other linux distros may work, but they need to be able to run the
apt
package manager) - A GPU with CUDA support (Nvidia GPUs are recommended)
- Training and Validation videos
- Clone this repository
git clone https://github.com/SeanErn/EasyTfodToolchain.git
- Enter the directory
cd EasyTfodToolchain
- Run the installer
python run.py
- Installer script fully working * Update 11/23/23: Working besides some issues with built in logger panic crashing textual when downloading CUDA via wget, tries to install cudNN when not fully downloaded
- Install error handling (report errors to user, suggest fixes)
- Python-wide logger
- Project Manager
- Find annotation software (maybe labelImg or LabelStudio) OR create from scratch D:
- Prepare, Train, Export python scripts
- Test python script
- FULL USER GUIDE
- MVP
- Advanced purge dataset, purge logs, reinstall, etc
- TF v1 support