search

JiehongLin / SAM-6D

[CVPR2024] Code for "SAM-6D: Segment Anything Model Meets Zero-Shot 6D Object Pose Estimation".
346 stars 30 forks source link

Problems encountered in pose inference process #30

Closed Jingranxia closed 8 months ago

Jingranxia commented 8 months ago

作者您好，我又遇到一个问题，我用自己的数据替换了demo的数据，在sam部分执行的非常好，然而在pose预测的过程中出现了问题，这个问题应该是什么问题呢？

load pre-trained checkpoint from: checkpoints/mae_pretrain_vit_base.pth => extracting templates ... => loading input data ... Traceback (most recent call last): File "/SAM-6D/SAM-6D/Pose_Estimation_Model/run_inference_custom.py", line 279, in input_data, img, whole_pts, model_points, detections = get_test_data( File "/SAM-6D/SAM-6D/Pose_Estimation_Model/run_inference_custom.py", line 246, in get_test_data ret_dict['pts'] = torch.stack(all_cloud).cuda() RuntimeError: stack expects a non-empty TensorList

Jingranxia commented 8 months ago

问题解决了，scale改成0.0001即可

kevinDrawn commented 8 months ago

product 1000 to your original depth scale.

or check your Model file unit (ply). It should be mm unit.