search

DengpanFu / LUPerson

Unsupervised Pre-training for Person Re-identification (LUPerson)
242 stars 35 forks source link

如何解析dets中的pkl #8

Open wang21jun opened 3 years ago

wang21jun commented 3 years ago

感谢您的工作，您dets中的pkl解析后每个item是如下所示格式，请问各个字段依此代表什么（比如1819代表？），另外这个检测信息如何知道对应到视频中的哪个帧？多谢． 1819: [[0, {'bbox': array([ 618.625, 81.75 , 1031.375, 709.25 ], dtype=float32), 'pose': array([[733.08105, 205.53418], [753.5075 , 176.93718], [720.8252 , 185.10774], [822.95734, 176.93718], [724.91046, 193.27832], [929.1748 , 328.09277], [716.7399 , 340.34863], [986.36884, 524.1865 ], [692.2282 , 462.90723], [790.2751 , 597.7217 ], [667.7165 , 528.2718 ], [855.63965, 695.76855], [716.7399 , 691.6833 ], [773.9339 , 720.2803 ], [761.67804, 691.6833 ], [896.4925 , 699.8538 ], [684.0576 , 703.93915]], dtype=float32), 'conf': 0.9488497, 'prob': array([[0.960111 ], [0.9625488 ], [0.96105295], [0.9494078 ], [0.5803897 ], [0.7805291 ], [0.8539633 ], [0.84015006], [0.9082806 ], [0.85463387], [0.9051482 ], [0.68101335], [0.77411926], [0.01756088], [0.01273601], [0.07441353], [0.05114838]], dtype=float32)}]]

ivskaikai commented 3 years ago

1819代表禎數的意思

wang21jun commented 3 years ago

非常感谢，我试一下．

shreejalt commented 3 years ago

@wang21jun Can you help me in how did you scraped the data from youtube using the given keys?

shreejalt commented 3 years ago

At what FPS youtube videos are read? It seems that when I parse the videos, random crops are obtained rather than original person crops? Do we need to change the fps of the youtube video when we read it?

flazerain commented 3 years ago

1819代表禎數的意思

请问一下视频的分辨率怎么确认呢？视频帧的bbox对应不上，而且每个视频正确的分辨率还不一样（不是最大分辨率）

shreejalt commented 3 years ago

@flazerain Yes the problem is of the resolution itself.

dongxuemin666 commented 3 years ago

Hi, congratulations for your excellent work. As I experimented, the "dets.pkl" you provided could not match the persons in the video. As above mentioned, this could be due to resolution difference. So could you please tell us how to solve this problem?

dongxuemin666 commented 3 years ago

@flazerain 请问您解决这个分辨率的问题了吗？

oliver8459 commented 2 years ago

@dongxuemin666 请问您解决分辨率问题了吗？求赐教！

zhixingc commented 2 years ago

前两天作者release相关的下载与解析代码，我们对应整理了一份数据集。百度云链接为： 链接：https://pan.baidu.com/s/1VPX5Cs0qYsT4CE5fD1b3Ug 提取码：LNsd 供大家参考！

wengdunfang commented 2 years ago

前两天作者release相关的下载与解析代码，我们对应整理了一份数据集。百度云链接为： 链接：https://pan.baidu.com/s/1VPX5Cs0qYsT4CE5fD1b3Ug 提取码：LNsd 供大家参考！

hi，您是怎么将data.mdb解析成图片的呢？

DengpanFu commented 2 years ago

前两天作者release相关的下载与解析代码，我们对应整理了一份数据集。百度云链接为： 链接：https://pan.baidu.com/s/1VPX5Cs0qYsT4CE5fD1b3Ug 提取码：LNsd 供大家参考！

hi，您是怎么将data.mdb解析成图片的呢？

data.mdb是lmdb数据的主文件，图片在里面是以{key:value}的形式存储的，要结合keys.pkl里面的key，并使用lmdb库去读取，可以参考lup_moco里dataset.py的方式去解析，也可以参考 https://github.com/DengpanFu/LUPerson/blob/fce9ba6c94b1199a98ada15a8a4ae89957e35dd0/LUP/convert_lmdb.py#L58-L64

Soar-Sir commented 2 years ago

前两天作者release相关的下载与解析代码，我们对应整理了一份数据集。百度云链接为： 链接：https://pan.baidu.com/s/1VPX5Cs0qYsT4CE5fD1b3Ug 提取码：LNsd 供大家参考！

Thank you very much for your work. Can you upload this dataset to AliDrive(https://www.aliyundrive.com)? Baidu disk has not opened a membership, the download speed is very slow...slow...slow...Thanks again!

Zhongdao commented 2 years ago

chengzhixing

Hi Zhixing, 感谢贡献！ 请问用您这个版本的数据可以大致复现到文章里的数字吗？

blue-blue272 commented 2 years ago

前两天作者release相关的下载与解析代码，我们对应整理了一份数据集。百度云链接为： 链接：https://pan.baidu.com/s/1VPX5Cs0qYsT4CE5fD1b3Ug 提取码：LNsd 供大家参考！

作者可以提供Google云的下载链接吗？下载下来是压错文件已损坏。

ycbilge commented 1 year ago

could you also provide the data using any other source like google drive? We cannot reach the Baidu disk.