Gul Varol

Sorry for the late reply. Yes, I converted C3D caffemodel into torch model (but not torch into caffe). [Here](https://github.com/gulvarol/ltc#4-c3d-model-in-torch), I added some information about how I did this. You can...

Have you created train/test folders for the data splits? You can find the split definitions here: http://crcv.ucf.edu/data/UCF101.php Find the link where it says: "The Train/Test Splits for Action Recognition on...

Hi, C1...Ck denote either two-channel motion (flow-x, flow-y) or three-channel appearance (R,G,B). I am afraid this question requires understanding basics of 3D convolutions. You might want to apply the model...

The temporal extent is simply the number of input frames (clip) of the network. `t ∈ {20,40,60,80,100}` is not on the `60f` network. It is either 20-frames (`20f`), 40-frames (`40f`),...

Yes 60f means 60 frames. The difference between 2 consecutive frames for 60f and 100f is the same since we always sample consecutive frames from the original video. More randomness...

This is explained in the last two paragraphs of Section 3.3 of the paper. At training, we take a random (not necessarily the first) 60-frame clip. At test time, we...

You may be right, I probably swapped by mistake h=-1 and h=3 while making the Figure A.4