We provide the training and evaluation codes for the downstream tasks which we have experimented on.
These tasks are outlined as follows.
Task name: path
, reference in the main paper
- object detection:
downstream/detection/
,Table 7c
- pixel-wise segmentation:
downstream/segmentation/
,Table 7c
- open-set semi-supervised learning:
downstream/opensemi/
,Table 6
- webly supervised learning:
downstream/weblysup/
,Table 6
- semi-supervised learning:
downstream/semisup/
,Table 15 in Appx. F.3
- active learning:
downstream/active/
,Table 16 in Appx. F.3
- hard negative mining:
downstream/mining/
,Table 17 in Appx. G
Note: Semi-supervised learning here includes MixMatch, ReMixMatch, FixMatch, and FlexMatch, all of which utilize both labeled and unlabeled target data. On the other hand, semi-supervised learning in train_sup.py utilizes only partially labeled data (Table 7b in the paper), which follows the same protocol as other SSL works.
Every running files are located in each task. For example, to run the OpenSemi experiment with OpenMatch method, run
$ cd downstream/opensemi/
$ bash run_openmatch.sh
All you have to do is setup the dataset and its path, and the pretrained model checkpoint.