abduallahmohamed/skeleton-graph

Code for: "Skeleton-Graph: Long-Term 3D Motion Prediction From 2D Observations Using Deep Spatio-Temporal Graph CNNs", ICCV2021 Workshops

skeleton-graph's Issues

The predicted future motion

Hi,

I saw that you predict the future 3d human poses in the current frame's camera coordinate. My question is that the predicted future 3D motion results shown in your paper are in the world coordinate, so we need additional camera future positions to transform this coordinate system. Actually, this transformation matrix or camera position information is unavailable for your method in practice. Hence, how can your method predict these future human motion in the 3D scene?

GTAIMFPS5

Hello,

I have a question regarding the line in the train.py script : datasections = glob.glob('./GTAIMFPS5/*/').

GTAIMFPS5 must contain what exactly please? can you tell me how many files in total and contain what please? because I have the database but I did not understand what exactly I should choose to put.

Sequence prediction

Hello,

I have a question about your approach.

How many sequences did you use for the observation and then for the prediction phase for GTA dataset please?

How many frames do you use to observe and then predict for how many frames?

Best regards,

Train.py issue

Hello,

when i run the train script i get this error :
File "train.py", line 93, in
loader_train = torch.utils.data.DataLoader(
File "C:\Anaconda3\envs\skeleton\lib\site-packages\torch\utils\data\dataloader.py", line 266, in init
sampler = RandomSampler(dataset, generator=generator) # type: ignore
File "C:\Anaconda3\envs\skeleton\lib\site-packages\torch\utils\data\sampler.py", line 103, in init
raise ValueError("num_samples should be a positive integer "
ValueError: num_samples should be a positive integer value, but got num_samples=0

How can I fix it please?

Thanks!

Question Architecture (images)

Hello,

Thank you for your amazing work! But I have a question. You use images in your architecture but in the code the configuration is set to False: Load_image, video_back and background_back are set set false in your code. How is this? Can you explain me please? Where is the part of the extract and use of images please? How you use your images please?

Thanks!

Visualization problem

Hello,

When I run the visualization script: CUDA_VISIBLE_DEVICES=1 python3 visualization.py --lr 0.01 --n_stgcnn 5 --n_txpcnn 11 --dataset GTA_IM --use_lrschd --num_epochs 450 --tag 1 --torso_joint 13 --eval_only

I keep getting this error that I don't understand!

Can you help me please?

Best regards,

Prox Problem

Hello,

I tried to test on Prox dataset after downloading the recoding folder. Please can you share me the recoding folder you have been working on? Because by launching the training, I cannot launch it, it tells me that frame x does not exist in the folder!

Thanks in advance

Camera to world coordinates

Hello, I need to make a figure as your figure 1 please how can I make this please!
from what I understood it must the world coordinates but the predictions of your model are coordinated images. How can I do this conversion exactly with the matrix please!

Loss function

Hello,

I don't understand well your loss function, Can you give me some details please! How you calculate the loss!

Another question: if I don't use the graph in my architecture, is it possible to use this same loss formula that you use please!

Thanks a lot!

Reproducing issue on PROX dataset

Hi,

I encountered an issue during the experiment on the PROX dataset. Despite following the parameters provided in the paper, I couldn't reproduce the results outlined in the study. Consequently, I am reaching out to request detailed parameter settings and have attached a screenshot showcasing my best attempt on the PROX dataset.

Thank you very much!

Torso Joint

Hello,
What is the torso_joint please and whats is its value in the PROX dataset !
Thanks