It can be frustrating to find that you have no access to a trained model. What make it worse are that the authors do not reply emails and their codes do not run properly. Therefore, I create this repository.
This repository contains the trained model(s) and details in training. The model is derived from the original UDH repository and I trained it on my own machine. Feel free to do anything as long as follow the license. Maybe in the future I will train and upload more of them.
Because of many limitations and barriers, I use split.py and add_txt.py to transfer model files. Use rm_txt.py and merge.py sequentially to recover them, you will then get a trained_model.tar.gz file.