search

UX-Decoder / Semantic-SAM

[ECCV 2024] Official implementation of the paper "Semantic-SAM: Segment and Recognize Anything at Any Granularity"
2.37k stars 117 forks source link

No module named 'MultiScaleDeformableAttention' #46

Open Chenjuanwen opened 1 year ago

Chenjuanwen commented 1 year ago

When I try to run any demo or eval python, I come across such problem：

ModuleNotFoundError: No module named 'MultiScaleDeformableAttention'

Please compile MultiScaleDeformableAttention CUDA op with the following commands: cd mask2former/modeling/pixel_decoder/ops sh make.sh

BUT there is no path mask2former/modeling/pixel_decoder/ops. How can I fix it ?

MaureenZOU commented 1 year ago

Please refer to https://github.com/facebookresearch/Mask2Former/tree/main/mask2former/modeling/pixel_decoder/ops

fefespn commented 1 year ago

@Chenjuanwen first I used pip env to install the libraries and I get the same issue as you 'ModuleNotFoundError: No module named 'MultiScaleDeformableAttention'' , when I used conda all things worked :D just create new conda env, and pip install inside that env, that what worked for me.

Liam-McKenna04 commented 1 year ago

for me running make.sh at semantic_sam/body/encoder/ops resolved this