Pingchuan Ma
Pingchuan Ma
It is still on the way. I will publish my results as soon as I make it.
@nemonameless Yes it should be pre-trained on Kinetics following the steps of original paper.
@Haijunlv Actually no. I am currently doing experiments using 8-frames. I'll try it soon thanku for the information.
I provided a demo with backbone of resnet in the repo. You can simply use it as a utility module using the code from TSN-PyTorch.
Sorry for the late reply! I just tried again deploying the environment from scratch and running the script you gave but it works fine on my side. Could you please...
This is a good question. MPM is a semi-particle-based method and thus there is not an underlying mesh. There are a few ways to simulate the effect though: 1. Reconstruct...
> We have a small confusion, we noticed that the volume of each particle is obtained by dividing the mesh volume by the number of particles. One issue is that...
> Is there any specific code to generate the textured videos? Just confused because not able to find method 2. in the code. The code for method 2 is not...
Sorry for the late reply! Please try to replace the `tape.py` file with [this file](https://github.com/PingchuanMa/SGA/blob/main/sga/warp/tape.py). The problem roots in the incompatibility of warp when upgrading from 0.6.1 to what you...
Thank you for the interest in our work. We have a list of hyperparameters we used to reproduce the previous works in the paper, simply altering them should give reasonable...