Skip to content

Commit 61f3b9c

Browse files
committed
adding generated motions from other methods, and explanation on how to evaluate on the readme
1 parent 2bd781c commit 61f3b9c

File tree

3 files changed

+36
-5
lines changed

3 files changed

+36
-5
lines changed

README.md

+18-4
Original file line numberDiff line numberDiff line change
@@ -41,7 +41,7 @@ conda create python=3.9 --name temos
4141
conda activate temos
4242
```
4343

44-
Install [PyTorch 1.10](https://pytorch.org/) inside the conda environnement, and install the following packages:
44+
Install [PyTorch 1.10](https://pytorch.org/) inside the conda environment, and install the following packages:
4545
```bash
4646
pip install pytorch_lightning --upgrade
4747
pip install torchmetrics==0.7
@@ -245,12 +245,26 @@ Currently, evaluation is only implemented on skeletons with [MMM](https://mmm.hu
245245

246246

247247
### Evaluating prior works
248-
WIP: the proper instructions and code will be available soon.
249248

250-
To give an overview:
249+
Please use this command line to download the motions generated from previous work:
250+
251+
```bash
252+
bash prepare/download_previous_works.sh
253+
```
254+
255+
Then, to evaluate a method, you can do for example:
256+
257+
```bash
258+
python evaluate.py folder=previous_work/ghosh
259+
```
260+
261+
or change "ghosh" with "jl2p" or "lin".
262+
263+
264+
To give an overview on how to extract their motions:
251265
1. Generate motions with their code (it is still in the rifke feature space)
252266
2. Save them in xyz format (I "hack" their render script, to save them in xyz npy format instead of rendering)
253-
3. Load them into the evaluation code (instead of loading TEMOS motions).
267+
3. Load them into the evaluation code, as shown above.
254268

255269
</details>
256270

prepare/download_pretrained_models.sh

+1-1
Original file line numberDiff line numberDiff line change
@@ -8,7 +8,7 @@ md5sum pretrained_models.tgz
88

99
echo -e "If it is not, please rerun this script"
1010

11-
sleep 3
11+
sleep 5
1212
tar xfzv pretrained_models.tgz
1313

1414
echo -e "Cleaning\n"

prepare/download_previous_works.sh

+17
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,17 @@
1+
#!/bin/bash
2+
echo -e "The generation of previous work (on the test set of KIT-ML) will stored in the 'previous_work' folder\n"
3+
gdown "https://drive.google.com/uc?id=1ZstP5VOyyKfqgec-G69SR4HEIvwWM5uc"
4+
5+
echo -e "Please check that the md5sum is: 6b753eec02b432448b70c72c131f0ae8"
6+
echo -e "+ md5sum previous_work.tgz"
7+
md5sum previous_work.tgz
8+
9+
echo -e "If it is not, please rerun this script"
10+
11+
sleep 5
12+
tar xfzv previous_work.tgz
13+
14+
echo -e "Cleaning\n"
15+
rm previous_work.tgz
16+
17+
echo -e "Downloading done!"

0 commit comments

Comments
 (0)