There is an error in the file about VRD dataset on Baidu cloud.

[YANYANYEAH]

Thank you very much for sharing. I downloaded the VRD dataset from the Baidu cloud link you provided. But I am having problems when I execute “run process/vrd_pred_process.py“

and i debug the process/vrd_pred_process.py Then I found a problem with the format of the json file. Then I downloaded the official website to provide download links and made a comparison. Found that your code works fine under the json file provided by the official website. This is the download link of the official website. https://cs.stanford.edu/people/ranjaykrishna/vrd/

[yangxuntu]

Thank you very much!

[xxxxuee]

+1

[PeterVennerstrom]

Using python 3.6 I get an error running vrd_pred_process.py at line 34.

TypeError: 'dict_keys' object does not support indexing

The error can be resolved by adding 'list' to line 27 like so:

image_name = list(data.keys())

Thanks!

[yangxuntu]

I use python 2.7 and you can use anaconda to set up one environment with python 2.7.

---

发件人: PeterVennerstrom notifications@github.com 发送时间: 2018年7月19日 22:36:19 收件人: yangxuntu/vtranse 抄送: #YANG XU#; Comment 主题: Re: [yangxuntu/vtranse] There is an error in the file about VRD dataset on Baidu cloud. (#3)

Using python 3.6 I get an error running vrd_pred_process.py at line 34.

TypeError: 'dict_keys' object does not support indexing

The error can be resolved by adding 'list' to line 27 like so:

image_name = list(data.keys())

Thanks!

― You are receiving this because you commented. Reply to this email directly, view it on GitHubhttps://github.com/yangxuntu/vtranse/issues/3#issuecomment-406298738, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AkiaOlqm3hVuFwC_SPfZJgF3q304hoqKks5uIJljgaJpZM4VO25N.

[cuienjiecv]

I have the same problem with him @PeterVennerstrom. I also get an error running vrd_pred_process.py and I use python 2.7 and tensorflow1.3. AttributeError Traceback (most recent call last) /home/cuienjie/vrd-master/process/vrd_pred_process.py in () 25 with open(file_path_use,'r') as f: 26 data=json.load(f) ---> 27 image_name = data.keys() 28 len_img = len(image_name) 29 t = 0

AttributeError: 'list' object has no attribute 'keys'

[cuienjiecv]

The sg_dataset.zip is downloaded from https://share.weiyun.com/55KK78Y. The dataset is list not dict. So I think it may be that the dataset has gone wrong.

[cuienjiecv]

@yangxuntu thank you very much the sg_dataset.zip which you provide download. the sg_train_annotations.json and sg_test_annotations.json are 'list' object has no attribute 'keys'. (For example : [{u'relationships': [{u'text': [u'car', u'has', u'front wheels'], u'objects': [0, 1], u'relationship': u'has'},.... {u'attributes': [], u'names': [u'shirt'], u'bbox': {u'y': 240.0, u'x': 734.0, u'w': 102.0, u'h': 147.0}}], u'height': 768}]) When I run script 'run process/vrd_pred_process.py' have an error.

/home/cuienjie/vrd/process/vrd_pred_process.py in () 25 with open(file_path_use,'r') as f: 26 data=json.load(f) ---> 27 image_name = data.keys() 28 len_img = len(image_name) 29 t = 0 AttributeError: 'list' object has no attribute 'keys'

Can you help me solve the mistake? Thank you very much!

[cuienjiecv]

The download link of the official website. https://cs.stanford.edu/people/ranjaykrishna/vrd/ But this link can't get in.

[yangxuntu]

You can go here to find the source data:https://github.com/zawlin/cvpr17_vtranse, and here to download the datahttps://drive.google.com/drive/folders/1BvtjCnlORMg4l92kNgZ2g1YaHYj9Dy3X.

[https://avatars3.githubusercontent.com/u/1491654?s=400&v=4]https://github.com/zawlin/cvpr17_vtranse

GitHub - zawlin/cvpr17_vtransehttps://github.com/zawlin/cvpr17_vtranse github.com GitHub is where people build software. More than 28 million people use GitHub to discover, fork, and contribute to over 85 million projects.

---

发件人: cuienjiecv notifications@github.com 发送时间: 2018年7月31日 16:37:05 收件人: yangxuntu/vrd 抄送: #YANG XU#; Mention 主题: Re: [yangxuntu/vrd] There is an error in the file about VRD dataset on Baidu cloud. (#3)

The download link of the official website. https://cs.stanford.edu/people/ranjaykrishna/vrd/ But this link can't get in.

― You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/yangxuntu/vrd/issues/3#issuecomment-409141954, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AkiaOovrsFn72kEWa9IEwh1IxxjrgkAHks5uMBcxgaJpZM4VO25N.

[yangshuo11]

Replace the two json files in sg_dataset (https://share.weiyun.com/55KK78Y) with the other two json files in https://drive.google.com/open?id=1BzP8DN2MAz76IvQTlpNOYla_bNC9gQuN. This could resolve the issue.

[onexs]

Whether the VG dataset after noise removal has a file package with a format similar to that in the VRD dataset folder

[emanuelazfernandes]

Original dataset Google Drive link.