Siyuan Yao
Results
3
comments of
Siyuan Yao
> Use cuda or remove # if FP16_INFERENCE # image = image.half() # model = model.half() in the grounding_dino_demo.py Works perfectly, thanks!
You can turn off rigid loss to speed up by 10x
> [@JCBreath](https://github.com/JCBreath) Can you explain in detail how to do it? Set lambda_rigid: 0.0 in your yaml config