Skip to content

Commit

Permalink
Update README.md
Browse files Browse the repository at this point in the history
  • Loading branch information
@wanglimin
wanglimin committed Mar 24, 2016
1 parent 3f91444 commit 4665e69
Showing 1 changed file with 4 additions and 4 deletions.
8 changes: 4 additions & 4 deletions README.md
View file
Expand Up @@ -12,10 +12,10 @@ Here we provide the code for the extraction of Trajectory-Pooled Deep-Convolutio
### Two-stream CNN models trained on the UCF101 dataset
First, we provide our trained two-stream CNN models on the split1 of UCF101 dataset, which achieve the recognition accuracy of 84.7%

["Spatial net model"](http://mmlab.siat.ac.cn/tdd/spatial.caffemodel) </br>
["Spatial net prototxt"](http://mmlab.siat.ac.cn/tdd/spatial_cls.prototxt) </br>
["Temporal net model"](http://mmlab.siat.ac.cn/tdd/temporal.caffemodel) </br>
["Temporal net prototxt"](http://mmlab.siat.ac.cn/tdd/temporal_cls.prototxt)
["Spatial net model (v1)"](http://mmlab.siat.ac.cn/tdd/spatial.caffemodel) </br>
["Spatial net prototxt (v1)"](http://mmlab.siat.ac.cn/tdd/spatial_cls.prototxt) </br>
["Temporal net model (v1)"](http://mmlab.siat.ac.cn/tdd/temporal.caffemodel) </br>
["Temporal net prototxt (v1)"](http://mmlab.siat.ac.cn/tdd/temporal_cls.prototxt)

### TDD demo code
Here, a matlab demo code for TDD extraction is provided.
Expand Down

2 comments on commit 4665e69

@lovebobo
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can I change the protoext input to [1,20,224,224]? when i do this,the feature of the optical flow is wrong。。。

@wanglimin
Copy link
Owner Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think it is OK to change the input to dim: [1, 20, 224, 224]. But at every time, you need to change the batch_size here accordingly:

https://github.com/wanglimin/TDD/blob/cudnn2.0/TemporalCNNFeature.m#L23

What do you mean by 'the feature of the optical flow is wrong'.

Please sign in to comment.