A toy implementation for article Yeh, Yu-Cheng, et al. "Deep learning approach for automatic landmark detection and alignment analysis in whole-spine lateral radiographs.". (WIP, it's kind of broken now.) Disclaimer: I'm not the author, so please don't contact them when encountering issues. Also, the repo is built for educational purpose, use with care.
first create a virtual environment and install the required packages, then run the train.py in that environment. That should do.