search

mks0601 / 3DMPPE_ROOTNET_RELEASE

Official PyTorch implementation of "Camera Distance-aware Top-down Approach for 3D Multi-person Pose Estimation from a Single RGB Image", ICCV 2019
MIT License
477 stars 65 forks source link

data/PW3D #26

Closed zqq-judy closed 3 years ago

zqq-judy commented 3 years ago

Hello！ I can't find this "joint_cam". And it is not found in the 3DPW_train.json file.

Traceback (most recent call last): File "train.py", line 83, in <module> main() File "train.py", line 33, in main trainer._make_batch_generator() File "/home/zqq/project/3DMPPE_ROOTNET_RELEASE-master/main/../common/base.py", line 94, in _make_batch_generator trainset3d_loader.append(DatasetLoader(eval(cfg.trainset_3d[i])("train"), True, transforms.Compose([\ File "/home/zqq/project/3DMPPE_ROOTNET_RELEASE-master/main/../data/PW3D/PW3D.py", line 18, in __init__ self.data = self.load_data() File "/home/zqq/project/3DMPPE_ROOTNET_RELEASE-master/main/../data/PW3D/PW3D.py", line 35, in load_data joint_cam = np.array(ann['joint_cam'], dtype=np.float32).reshape(-1,3) KeyError: 'joint_cam'

mks0601 commented 3 years ago

Sorry I updated .json files. Now, it will work.

zqq-judy commented 3 years ago

Hi, I have another question. In the xxx.py file of the loaded datasets, there is a line of code. For example, in MSCOCO.py, joint_img = np.array(ann['keypoints']).reshape(-1,3) , what is the meaning of keypoints ? What is the purpose of this line of code?

mks0601 commented 3 years ago

It is a set of 2D joint coordinates in MSCOCO. See here