search

TencentYoutuResearch / ActionDetection-AFSD

Code for CVPR2021 paper "Learning Salient Boundary Feature for Anchor-free Temporal Action Localization"
Other
171 stars 31 forks source link

ActivityNet v1.3 data preprocess & infernece #15

Open fanqing-1116 opened 3 years ago

fanqing-1116 commented 3 years ago

I downloaded all the sampled video data（32.4G), the total number of these videos is 14950. But the total number of all npy files I get after running step 3 is only 11171. When I run the RGB model inference I also get some FileNotFoundError like "No such file or directory: 'datasets/activitynet/train_val_npy_112/v_JDg--pjY5gg.npy'". I wish i can use some help.

yaru-zhang commented 3 years ago

@linchuming Hi, thanks for your work! I also met with the same problem when I preprocess the ActivityNet dataset. Specifically, when the duration time of a video is a little bit longer, I can not extract RGB and flow npy features. Would you mind giving me some help to figure it out.

linchuming commented 3 years ago

@fanqing-1116 , did you try use 7za x -r train_val_112.7z and check the number of mp4 files is 14950?

czhaneva commented 3 years ago

Why the number of classes in ActivityNet-1.3 is set to 2 instead of 201. The paper does not mention it. So I am a little curious that it is a action proposal task in ActivityNet-1.3. Hope to get your reply, thank you~

linchuming commented 3 years ago

@czhaneva you can refer to the answer of the issue #4 .

Liu-arch commented 1 month ago

@linchuming Hi, thanks for your work! I also met with the same problem when I preprocess the ActivityNet dataset. Specifically, when the duration time of a video is a little bit longer, I can not extract RGB and flow npy features. Would you mind giving me some help to figure it out.

Hello, activitynet v1.3 can't download the data, can you itemize the data?