Warning
This repogitory is work in progress.
This repository converts the XMem tracker model included in Track-Anything from PyTorch to ONNX.
name | version |
---|---|
os | windows 10 |
cuda | 11.8 |
python | 3.10.15 |
uv | 0.4.20 |
torch | 2.3.1 |
torchvision | 0.18.1 |
sample mask file: test-sample1-1frame-mask.png
is made by EfficientSAM
git clone https://github.com/5PB-3-4/XMem_Export.git
Check out requirement.txt.
Original pretrained model is here.
Tested XMem-s012.pth, XMem-with-mose.pth
# Run
cd XMem_Export
python export.py -i ./ckpt/XMem-s012.pth --width 640 --height 480 --mask_num 1
# Parser option
python export.py -h