This research project implements a real-time object detection and pose estimation method as described in the paper, Tekin et al. "Real-Time Seamless Single Shot 6D Object Pose Prediction", CVPR 2018. (https://arxiv.org/abs/1711.08848).
I notice that ,there is code in train.py :
bg_file_names = get_all_files('VOCdevkit/VOC2012/JPEGImages')
train_loader = torch.utils.data.DataLoader(bg_file_names=bg_file_names。。。。。)
I wonder what is the role of voc2012 when training my own dataset
They are used for data augmentation, used to replace the background of an object of interest. In case you put False for data augmentation, they are not used, you can ignore.
I notice that ,there is code in train.py : bg_file_names = get_all_files('VOCdevkit/VOC2012/JPEGImages') train_loader = torch.utils.data.DataLoader(bg_file_names=bg_file_names。。。。。)
I wonder what is the role of voc2012 when training my own dataset