XMEM V7 Fork

To launch training: python -m torch.distributed.launch --master_port 25763 --nproc_per_node=2 train.py --exp_id EXPERIMENT_NAME --stage 3 --load_network saves/XMem-s012.pth

Training data

Original training data has been moved to Dodo at /data/thom/xmem_training_data You configure the dataloaders in train.py in the functions named renew_<datasetname>_loader The format for each "video" is a folder of Annotations (.png images, 2D, 0,1) and a folder of images. For in context learning you would just have 1 video.

Name		Name	Last commit message	Last commit date
Latest commit History 94 Commits
dataset		dataset
docs		docs
inference		inference
model		model
scripts		scripts
util		util
.gitignore		.gitignore
Dockerfile		Dockerfile
LICENSE		LICENSE
Makefile		Makefile
ORIGINAL_README.md		ORIGINAL_README.md
README.md		README.md
dataset_load_test.py		dataset_load_test.py
eval.py		eval.py
interactive_demo.py		interactive_demo.py
merge_multi_scale.py		merge_multi_scale.py
requirements.txt		requirements.txt
requirements_demo.txt		requirements_demo.txt
train.py		train.py

License

v7labs/XMem

Folders and files

Latest commit

History

Repository files navigation

XMEM V7 Fork

Training data

About

Resources

License

Stars

Watchers

Forks

Languages