search

Petrichor625 / BATraj-Behavior-aware-Model

[AAAI 2024] Official PyTorch Implementation of ''BAT: Behavior-Aware Human-Like Trajectory Prediction for Autonomous Driving''.
https://arxiv.org/abs/2312.06371
79 stars 8 forks source link

关于dataset和utils.py #6

Open Zhangxx1218 opened 7 months ago

Zhangxx1218 commented 7 months ago

I used the NGSIM dataset provided by the author for training, but unfortunately encountered the following error. Additionally, the shape of the adjacency matrix is [41,41], which differs from the author's setting of 39. This seems to be a problem caused by the dataset. image And, when I trained with the raw dataset provided by CS-LSTM, I could generate checkpoints. However, during validation, due to the lack of the correct format for datasets like "keep" and "merge", I only used the original test set provided by CS-LSTM for overall testing. The experimental results are as follows, unable to reproduce the results of the original paper. Could the author provide some clarification on this? image

yyn010419 commented 7 months ago

@Zhangxx1218 hello, have you solve this problem(41 39)? I have met the same problem with u

Zhangxx1218 commented 7 months ago

@Zhangxx1218你好，这个问题你解决了吗（41 39）？我遇到了和你同样的问题

No.But I find another dataset that processed by NGSIM,it can train successfully!

yyn010419 commented 7 months ago

@Zhangxx1218 Could you please share it with me ? thanks a lot!!! If could, my email is yyn010419@126.com thanks again!

ontoon commented 7 months ago

@Zhangxx1218

Could you share it with me too?I met the same problem.thank you from the bottom of my heart.my email is 1574529553@qq.com.thanks.

guoyage commented 7 months ago

I used the NGSIM dataset provided by the author for training, but unfortunately encountered the following error. Additionally, the shape of the adjacency matrix is [41,41], which differs from the author's setting of 39. This seems to be a problem caused by the dataset. image And, when I trained with the raw dataset provided by CS-LSTM, I could generate checkpoints. However, during validation, due to the lack of the correct format for datasets like "keep" and "merge", I only used the original test set provided by CS-LSTM for overall testing. The experimental results are as follows, unable to reproduce the results of the original paper. Could the author provide some clarification on this? image

Have you reproduced the results of the NGSIM data set in the paper? The error of my own training model on the test set is slightly larger than the results shown in the author's paper.

yyn010419 commented 7 months ago

@guoyage could you share your dataset with me ?thank you a lot!

Zhangxx1218 commented 7 months ago

I used the NGSIM dataset provided by the author for training, but unfortunately encountered the following error. Additionally, the shape of the adjacency matrix is [41,41], which differs from the author's setting of 39. This seems to be a problem caused by the dataset. image And, when I trained with the raw dataset provided by CS-LSTM, I could generate checkpoints. However, during validation, due to the lack of the correct format for datasets like "keep" and "merge", I only used the original test set provided by CS-LSTM for overall testing. The experimental results are as follows, unable to reproduce the results of the original paper. Could the author provide some clarification on this? image

Have you reproduced the results of the NGSIM data set in the paper? The error of my own training model on the test set is slightly larger than the results shown in the author's paper.

Me too.The RMSE that I trained is larger than what the paper writed.

guoyage commented 7 months ago

I used the NGSIM dataset provided by the author for training, but unfortunately encountered the following error. Additionally, the shape of the adjacency matrix is [41,41], which differs from the author's setting of 39. This seems to be a problem caused by the dataset. image And, when I trained with the raw dataset provided by CS-LSTM, I could generate checkpoints. However, during validation, due to the lack of the correct format for datasets like "keep" and "merge", I only used the original test set provided by CS-LSTM for overall testing. The experimental results are as follows, unable to reproduce the results of the original paper. Could the author provide some clarification on this? image

Have you reproduced the results of the NGSIM data set in the paper? The error of my own training model on the test set is slightly larger than the results shown in the author's paper.

Me too.The RMSE that I trained is larger than what the paper writed.

This is my QQ1373976150. You can add contact information to communicate.

caoxu0109 commented 7 months ago

@Zhangxx1218 Could you share it with me too?I met the same problem.thank you from the bottom of my heart.my email is 2837128824@qq.com.thanks.

Petrichor625 commented 6 months ago

Dear all, Thanks for your interest in our work!

We conducted this research on a separate server, and the code uploaded is from an early version. The version provided here doesn't match the dataset used in our latest work. Unfortunately, we can't provide the latest version of the code immediately. Your patience is appreciated as we work to update with the latest code and dataset. Thank you again for your patience and understanding.

Alexsupreme5 commented 5 months ago

@Zhangxx1218 Could you share it with me too? I met the same problem. thank you from the bottom of my heart. my email is 75467991@.com. thanks again!

Curtainzsl commented 2 months ago

@Zhangxx1218 Could you share it with me too? I met the same problem. thank you from the bottom of my heart. my email is [799339017@qq.com] thanks again!

HHxxwshTTxs commented 2 months ago

我使用了笔者提供的 NGSIM 数据集进行训练，但很遗憾遇到了以下错误。此外，邻接矩阵的形状为 [41,41]，这与作者的 39 不同。这似乎是数据集引起的问题。 图像而且，当我使用 CS-LSTM 提供的原始数据集进行训练时，我可以生成检查点。但是，在验证过程中，由于缺少 “keep” 和 “merge” 等数据集的正确格式，我只使用了 CS-LSTM 提供的原始测试集进行整体测试。实验结果如下，无法重现原始论文的结果。作者能否对此进行一些澄清？ 图像

Can you please share the preprocessing file for the NGSIM dataset, I noticed that he has been removed from the project Releases list, thanks a lot! My email is [WangShihao@emails.bjut.edu.cn] thanks again!

HHxxwshTTxs commented 2 months ago

我使用了笔者提供的 NGSIM 数据集进行训练，但很遗憾遇到了以下错误。此外，邻接矩阵的形状为 [41,41]，这与作者的 39 不同。这似乎是数据集引起的问题。图像而且，当我使用 CS-LSTM 提供的原始数据集进行训练时，我可以生成检查点。但是，在验证过程中，由于缺少 “keep” 和 “merge” 等数据集的正确格式，我只使用了 CS-LSTM 提供的原始测试集进行整体测试。实验结果如下，无法重现原始论文的结果。作者能否对此进行一些澄清？image

您是否在论文中重现了 NGSIM 数据集的结果？我自己的训练模型在测试集上的误差比作者论文中显示的结果略大。

我也是。我训练的 RMSE 比论文写的要大。

这是我的QQ1373976150。您可以添加联系信息进行通信。

Can you please share the preprocessing file for the NGSIM dataset, I noticed that he has been removed from the project Releases list, thanks a lot! My email is [WangShihao@emails.bjut.edu.cn] thanks again!

AWAKENGKX commented 4 weeks ago

If possible, could I share some relevant questions with you? This is my QQ: 1449235421, thank you very much!

Petrichor625 commented 1 week ago

Thank you all for your interest in our paper, and I apologize for the delayed response.

At the moment, there is an issue with the version of the code currently available in this repository, and unfortunately, we’re unable to upload the latest corrected version right now. However, for those looking to replicate or improve upon the results, I recommend exploring our other project, HLTP (GitHub Link), which provides more effective methods and enhanced code that may be beneficial to your work.

If you have any questions or run into challenges with HLTP, please don’t hesitate to reach out! As I may not always receive notifications from GitHub or have the chance to check issues regularly, please feel free to contact me via email if you need a quicker response. I’ll do my best to assist you as soon as possible.

Thank you again for your patience, understanding, and support—it’s truly appreciated!