Hi,
I've created a pascal voc like data set with .xml annotation files and .txt for training and trainval.
I got two questions and I will very much appreciate your response:
1)Is there a way to only use my own data set, and if so which python files should I edit?
2)if the trainval.txt is used for evaluating the training what is the need for val.txt? why do we use test.txt? thanks!
1) You can create a new data file by imitating data/voc0712.py
2) val.txt if for evaluating the model trained on train.txt and test.txt is for the model trained on trainval.txt
Most helpful comment
1) You can create a new data file by imitating data/voc0712.py
2) val.txt if for evaluating the model trained on train.txt and test.txt is for the model trained on trainval.txt