hassony2/kinetics_i3d_pytorch
Inflated i3d network with inception backbone, weights transfered from tensorflow
PythonMIT
Issues
- 0
Grayscale Images
#34 opened by mnik17 - 0
Moving model weight file to some storage
#33 opened by priteshgohil - 1
The padding of I3D model should be symmetrical
#32 opened by fjchange - 0
- 2
convert kinetics-600 model failed
#26 opened by FingerRec - 4
Details regarding optical flow pre-processing
#19 opened by simonguiroy - 3
about extract features from my dataset
#30 opened by galaxysan - 1
could i ask the question on 2d conv?
#29 opened by Zhang-O - 1
Is the model a 16 frames/clip model?
#28 opened by KT27-A - 1
How can i use the code to train a network?
#24 opened by Epiphqny - 1
- 5
- 0
extract features from small videos
#23 opened by devraj89 - 0
Can anyone reach the result on UCF101? 95.6? I can't reach, l want to know more details!
#21 opened by tanxjtu - 1
- 2
python i3d_tf_to_pt.py --rgb
#17 opened by Zhuysheng - 1
Fine-tune on HMDB51 or other datasets
#16 opened by LiliMeng - 1
Pretrained model on Kinetics
#15 opened by LiliMeng - 1
accuary on UCF101 or Kinetics
#14 opened by Kathrine94 - 3
Extract video features?
#10 opened by gkv91 - 1
Handling videos at higher res
#12 opened by siddhantjain - 1
About data preprocessing
#13 opened by TianjiPang - 5
- 3
Transfer Learning
#8 opened by vateye - 1
Dropout: opposite meaning?
#6 opened by seann999 - 3
I think it should be ceil_mode=True
#4 opened by rimchang - 1
why use Ugly hack?
#3 opened by rimchang