Comments (9)
Hi! For the model we have adopted the following codebases https://github.com/papagina/Auto_Conditioned_RNN_motion and https://github.com/facebookresearch/QuaterNet/tree/main. Which json are you referring to?
from poserepresentation.
from poserepresentation.
from poserepresentation.
- Different json files correspond to different representations, loss weights and so on that you can build depending on the model you want to test out.
- The template skeleton is not SMPLH. We are using data from the CMU dataset, therefore we use the template bvh from there. You can find more information about which subjects we use exactly in the paper.
- You can extract the 3D position from the dual-quaternions (with current coordinate system) using this function.
You might have to restructure the eval.py file so that you can use the functions from dualquats by replacing
from code1.dualquats import currentdq2localquats
with from dualquats import currentdq2localquats
. You will have to also define your dataloader and network init functions (I can help you on that). In this codebase, we just provide the tools to convert between the representations.
from poserepresentation.
from poserepresentation.
- You can acquire the dataset from here. The json file will depend on the dataloader and model initiator functions that you will define. I can share with you the ones I used, but you will have to modify it depending on your functions.
- Yes, input and target must be the same.
- Potentially yes. Please look into this.
from poserepresentation.
from poserepresentation.
My dataset is the exact sample of bvh files that I sent you. I will try to provide you with a json later today. The libraries that relate to model and data loading you will have to define them yourself.
from poserepresentation.
A sample config file looks like this:
{ "visdom_title": "v1-test",
"model": "dq_acLSTM",
"acRNN_model": "vanilla",
"data_type": "dualquats",
"paired_dir": <train_data_path>,
"test_paired_dir": <test_data_path>,
"validation_paired_dir": <val_data_path>,
"order": "zyx",
"fps_current":120,
"fps_target":30,
"QRL_local": false,
"QRL_curr": false,
"DL": true,
"BL": false,
"ortho6D_loss": false,
"mse_loss":true,
"push_seq":"86_06",
"save_bvh_dir": <path_to_save_bvh>,
"save_weights_dir": <path_for_net_weights_save>,
"train_read_weight_path": "",
"bvh_skeleton_file_name": <default_skeleton_path>,
"save_weights_every": 10000,
"save_bvh_every": 1000,
"norm_type": "all",
"foot_contact": false,
"test_generate_seq_num": 3000,
"test_save_bvh_dir": <path_to_save_bvh_test>,
"test_read_weight_path": <path_for_net_weights_load>
}
As I said before, this is specific to our network initialization function.
from poserepresentation.
Related Issues (1)
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from poserepresentation.