HERIUN

```python from torchvision.models import googlenet import torch model = googlenet(pretrained=True) extractor = torch.nn.Sequential(*list(model.children())[:-2]) im = torch.randn(1,3,720,1280) # NCHW feature = extractor(im).cpu().numpy().flatten() # [1,1024,1,1] -> [1024] ``` i try like this...

저도 님 말씀이 맞는거 같습니다.

저도 확실하진 않은데, 현재 쓰는 PC OS가 window이면, docker에서 gpu를 사용하는것은 불가능한걸로 알고 있습니다. 우분투면 가능한거 같아요.

@ruanzhijian It may help you. https://github.com/HERIUN/vsumm-reinforce_re/blob/main/generate_dataset.py

```python import h5py datasets_ = h5py.File('datasets/eccv16_dataset_summe_google_pool5.h5') print(datasets_['video_1']['video_name'][...]) ```

Hi. I'm also trying to reproduce feature of vsumm video. but failed.. My all process steps are ```python import torch from torchvision.models import googlenet self.device = torch.device('cuda:0') self.googlenet = googlenet(pretrained=True)...

There's no CNN encoder. because dataset(.h5) already extracted..

I share with my google drive yolor_csp_x.pt : (https://drive.google.com/file/d/14gXs9bBHd8cCCIUHCkXIvmS2TSimZFAb/view?usp=sharing) yolor_csp_x_star.pt : https://drive.google.com/file/d/1QyJFriCQF-5da1PW12_tvr9qfWHvqnHN/view?usp=sharing