ROMP

ROMP copied to clipboard

Distributed training raise an error

8

Hi, I am trying to run Romp in distributed mode. I follow this [Script](https://github.com/Arthur151/ROMP/blob/master/scripts/train_distributed.sh). Since there is no folder called `core` in the repository I replaced it with `romp`. However,...

mkhoshle

Fix the instantiation of Visualizer in romp.lib.dataset.image_base.test_dataset()

JinkaiZheng

Hello. I am running your demo webcam.sh and I generated a cartoon character. I wonder are there any apis for changing the background?

NoLookDefense

[Config] Introduce calibration option

1

I have run some tests with different dancing video materials. Sometimes it recognizes the wrong body size at some frames, because of difficult movements. That doesn't look pretty. So how...

vivi90

How can I change the cartoon character?

8

Hello. Thanks for your work. I successfully ran your demo webcam.sh, which generates a cartoon character like this. I wonder how to generate such a character, and can I change...

CheungBH

小模型训练到30epoch loss值不下降，infer视频之后发现SMPL body始终在人物上方

17

[V1_mbv3_h36m,mpiinf,coco,mpii,lsp,muco,crowdpose_g0,1,2,3.log](https://github.com/Arthur151/ROMP/files/8338065/V1_mbv3_h36m.mpiinf.coco.mpii.lsp.muco.crowdpose_g0.1.2.3.log) 可以帮忙看一下这个mobilenetv3的训练日志吗？ 他的validation 指标下降到140&90附近就感觉卡住了，而且我把30epoch的模型拿出来测试了一下，发现SMPL body会飘在人物上面，我的想法是训练的时候body center heatmap应该是最先收敛的，所以即便pose上误差比较大，det loss应该收敛，也就是不会产生这种飘在人物上方的现象，不知道您训练的时候有没有这种问题，希望您能给出建议～ 我在resnet的基础上将batch_size 调整为了128，lr 相应x2 为 0.0001 图示：

hitsz-zuoqi

How to use ROMP to achieve motion capture with webcam

6

Here is the windows 10 platform, I see a video link on readme file shows ROMP can be used to achieve multi-person 3D mesh recovery in real time with webcam...

QiFuChina

How to value？

2

你好， import numpy as np import torch test = np.load('./00000000.npz',allow_pickle=True) print(test.files) print(test['results']['smpl_thetas'])我想用这个取值，为何报了这个错 IndexError: only integers, slices (`:`), ellipsis (`...`), numpy.newaxis (`None`) and integer or boolean arrays are valid indices

QingLuogwj

Is it possible to continue learning from any epoch?

1

Hey! Thanks for your work! I have a question: can I continue training the model from the epoch before which the model was trained? I mean, I started training from...

nikkorejz

key item

1

I try to run the code, but when i want to run smpl_parser there're no key values "outputs['smpl_betas'], outputs['smpl_thetas']",do I miss somthing? thanks for the answer

namachapi