wangxiang1230 / SSTAP

Code for our CVPR 2021 Paper "Self-Supervised Learning for Semi-Supervised Temporal Action Proposal".
71 stars 8 forks source link

about i3d feature #9

Open menghuaa opened 2 years ago

menghuaa commented 2 years ago

Hello, before linear interpolation, is the feature of I3D extracted at the original frame rate or at the frame rate of 25FPS? 您好,请问在没有线性插值前,I3D的特征是以原始帧频提取的,还是以25fps的帧频提取的?

wangxiang1230 commented 2 years ago

I3D的特征是以原始帧频提取的

Hi, the I3D features are extracted at the original frame rate.

menghuaa commented 2 years ago

I3D的特征是以原始帧频提取的

Hi, the I3D features are extracted at the original frame rate.

那请问您,是您自己提的特征吗?

menghuaa commented 2 years ago

I3D的特征是以原始帧频提取的

Hi, the I3D features are extracted at the original frame rate.

i3d的特征是RGB和光流特征拼接后进行线性插值的吗?

wangxiang1230 commented 2 years ago

I3D的特征是以原始帧频提取的

Hi, the I3D features are extracted at the original frame rate.

i3d的特征是RGB和光流特征拼接后进行线性插值的吗?

The I3D features provided only include RGB features and do not involve Flow.

menghuaa commented 2 years ago

I3D的特征是以原始帧频提取的

Hi, the I3D features are extracted at the original frame rate.

i3d的特征是RGB和光流特征拼接后进行线性插值的吗?

The I3D features provided only include RGB features and do not involve Flow.

我好像在论文中没有看到您使用i3d特征进行提名生成的结果,请问您使用该i3d特征的提名生成效果怎么样呢?与原始BMN相比

wangxiang1230 commented 2 years ago

I3D的特征是以原始帧频提取的

Hi, the I3D features are extracted at the original frame rate.

i3d的特征是RGB和光流特征拼接后进行线性插值的吗?

The I3D features provided only include RGB features and do not involve Flow.

我好像在论文中没有看到您使用i3d特征进行提名生成的结果,请问您使用该i3d特征的提名生成效果怎么样呢?与原始BMN相比

image In Table 1(b), we compared the performance differences between SSTAP and BMN.

wangxiang1230 commented 2 years ago

I3D features do not involve label information in the ActivityNet dataset (v1.2 or v1.3) and only use pre-trained Kinetics information.

menghuaa commented 2 years ago

I3D features do not involve label information in the ActivityNet dataset (v1.2 or v1.3) and only use pre-trained Kinetics information.

请问您是如何将thumos14的数据集划分为有标签数据集和无标签数据集的?我看到thumos14里每个视频的类别数量还是挺多的,不像activitynet1.3那样单一