Closed sinanmut closed 1 month ago
Hi, thanks for your interest in our work. The running time of our model is evaluated by online inferring it on a few motion sequences from the test set. We reported the average per-frame speed here. Please refer to our inference code about this.
You can also refer to this issue
Hi, thanks for sharing the code. Results look really good. I actually have a question about the inference time. It is reported in the paper that running time is 0.74ms per frame on NVIDIA RTX3090. Is it that time actually for the Refiner network? or end2end running per frame?
Thanks a lot.