Xiaolong Liu

@takfate @0829MingHu I have sent the code via email to you.

It seems that there's something wrong with your dataloader. No ground truth is found during mAP calculation. Does your dataset only contain one action class? Is your annotation file in...

Please refer to README. I explained it in the last paragraph of section 3 (Training by Yourself).

First, the results fluctuate much because THUMOS14 is a small dataset. Second, `disable_cuda=True` will also disable the actionness regression module, which will lead to performance drop. Last, there is a...

Hi, I think the results are close to those reported in the paper. Xiaolong 在 Kaan Akan ***@***.***>，2023年2月23日 15:45写道： Hello again, thank you for your response. I compiled the roi...

@pgupta119 You may refer to this page https://github.com/xlliu7/E2E-TAD/blob/master/docs/1_train_on_your_dataset.md

Thanks for your interest. Actually, this codebase already includes the training code. The code released here is re-written based on my internal implementation. I doubted that there might be bugs...

I guess it is okay after some modification. Note that torchvision.ops.RoIAlign only works for 2D images (4D feature maps of shape NxCxHxW). So you should resize the input feature sequence...

你好，这些是特征文件。首先将视频解码，然后使用预训练的特征提取器提取特征，最后保存到独立文件中。为了方便复现，我们直接提供了特征文件。详情可以参阅论文Implementation Details 获取Outlook for Android ________________________________ From: Hwa1998 ***@***.***> Sent: Tuesday, December 16, 2025 12:21:10 PM To: xlliu7/TadTR ***@***.***> Cc: Subscribed ***@***.***> Subject: [xlliu7/TadTR] 关于thumos14数据集 (Issue #29) [https://avatars.githubusercontent.com/u/157572072?s=20&v=4]Hwa1998 created an...