The problem of data - Githubissues

zuoym15 / craves.ai

CRAVES: Controlling Robotic Arm with a Vision-based, Economic System

https://craves.ai

GNU General Public License v3.0

261 stars 79 forks source link

The problem of data #5

Open Zhao-zw opened 5 years ago

Zhao-zw commented 5 years ago

The data ft_20181105/d3_preds has 2786 items but ft_20181105/imgs ft_20181105/imgs and ft_20181105/visualization has 2856 items. and also, the four data-zip always extract failed. Hope you renew the data. Thank you !

zuoym15 commented 5 years ago

Hi, thank you for your comment. The number of files in "d3_preds" and "imgs" are inconsistent because we filtered out images with too few valid keypoints or the 3D reconstruction error (i.e. formula #3 in our paper) is too high (>15), as these noisy predictions may harm the training process. The images filtered out are just ignored when loading the training data. As for the zip-file, we compressed it with WinRAR and we don't face the extract problem. Please provide more information on the platform you use if you still cannot extract them so that we can try to reproduce the bug.

wangjeidong commented 4 years ago

The data ft_20181105/d3_preds has 2786 items but ft_20181105/imgs ft_20181105/imgs and ft_20181105/visualization has 2856 items. and also, the four data-zip always extract failed. Hope you renew the data. Thank you !

I am trying to reproduce this paper and learn from it, but the process is not smooth. Run train_ arm.py , an error occurred. The link is a screenshot. I think it may be my dataset file path error, but I'm not sure. So, I want to know how to set the dataset path and program path with you. Looking forward to receiving your reply, thank you!