Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Issue: Serious Image Misalignment in SSCBench-nuScenes Dataset 图像文件数据错乱 #26

Open
NickHezhuolin opened this issue Mar 14, 2025 · 1 comment

Comments

@NickHezhuolin
Copy link

Hello,
I found a serious misalignment issue in the SSCBench-NUScenes dataset under the path:
📂 SSCBench/sscbench-nuscenes/trainval/sequences/

  1. The scene sequences in sscbench-nuscenes not match the nuScenes dataset correctly.
    
  2. The number of images in each scene is not consistent with nuScenes.
    
  3. There are images from different scenes should mixed together.
    

The scene sequences do not align correctly with the original nuScenes sequences. The only properly aligned sequences are 000000 and 000001, while the rest are mismatched.

Specific Issues
In the nuScenes dataset, scene 000 contains only 40 images ( you can check with code below).
📌 However, in SSCBench, the folder: SSCBench/sscbench-nuscenes/trainval/sequences/000000/image_2/
contains 41 images (0040.png exists but shouldn't).

The last image (0040.png) in 000000/image_2/ actually belongs to scene 001.This means scene transitions are incorrectly handled in SSCBench.
The dataset only contains sequences 000000 ~ 000698, meaning some sequences from nuScenes might be missing or misaligned.

Could you please recheck the dataset alignment and provide an update? This issue could significantly impact model training and evaluation.

Thank you for your work! 🚀

@NickHezhuolin
Copy link
Author

你好,

我在 SSCBench-NUScenes 数据集的以下路径中发现了 严重的对齐问题:
📂 SSCBench/sscbench-nuscenes/trainval/sequences/
问题概述

sscbench-nuscenes 中的场景序列与 nuScenes 数据集的序列不匹配。
每个场景的图像数量与 nuScenes 数据集不一致。
不同场景的图像被混合在一起,导致数据错误。

在 sscbench-nuscenes 数据集中,只有 前面第一个序列是部分对齐的,其余的序列都存在错位问题。

比如说:
在 nuScenes 数据集中,场景 000 仅包含 40 张图像(可以使用代码进行验证)。
📌 但是,在 SSCBench 数据集中,文件夹:
📂 SSCBench/sscbench-nuscenes/trainval/sequences/000000/image_2/
却包含 41 张图像(0040.png 存在,但它本不应该在这里)。

实际上,000000/image_2/0040.png 这张图片属于场景 001,说明 SSCBench 数据集在场景切换时存在错误。
此外,该数据集仅包含 序列 000000 ~ 000698,意味着可能有一些 nuScenes 的场景丢失或错位。

能否请您重新检查数据集的 场景序列对齐情况 并提供更新?
这个问题可能会 严重影响模型的训练和评估。

感谢您的工作!🚀

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant