About the training iterations of supervised model and semipervised model

facebookresearch / unbiased-teacher

PyTorch code for ICLR 2021 paper Unbiased Teacher for Semi-Supervised Object Detection

https://arxiv.org/abs/2102.09480

MIT License

409 stars 84 forks source link

About the training iterations of supervised model and semipervised model #48

Open Maggie-000 opened 2 years ago

Maggie-000 commented 2 years ago

Hello! I am confused about the training iterations of supervised model and its corresponding semi superviesed model. I want to know if the config ("faster_rcnn_R_50_FPN_focal_cross_coco_unsupW2.yaml") is the config of 3x experiments of coco + coco additinal datesets. The 3x means 3 times of labeled data or 3 times of all used data? Can you descripe the steps of Supervised(3x) experiments in Table2 of paper?