I test the work on the linemod dataset and the YCB dataset. It plays as well as the author mentioned in the paper on the Linemod, however, the ADD-S only 52.4 when I follow the advice of the author and train the YCB dataset. I have not changed the train-ycb.py and train-ycb.sh. Meanwhile, I test the author's pre-trained model and it works well, what is the problem?
I test the work on the linemod dataset and the YCB dataset. It plays as well as the author mentioned in the paper on the Linemod, however, the ADD-S only 52.4 when I follow the advice of the author and train the YCB dataset. I have not changed the train-ycb.py and train-ycb.sh. Meanwhile, I test the author's pre-trained model and it works well, what is the problem?