google-deepmind/kinetics-i3d
Convolutional neural network model for video classification trained on the Kinetics dataset.
PythonApache-2.0
Issues
- 1
- 2
How to create a custom action recognition model
#108 opened by snehitvaddi - 11
Does the video need to be cropped?
#119 opened by mazatov - 0
- 0
Run time of I3D on edge decives
#124 opened by 21kc-caracol - 2
Struggling to learn using Opt. Flow
#122 opened by mazatov - 5
- 1
Calculation of TV L1 flow
#120 opened by elpabos - 3
rgb.npy and flow.npy
#114 opened by niu1970 - 2
training from scratch
#115 opened by Tortoise17 - 2
- 0
Incompatibility issues
#123 opened by camruggles - 5
Optical flow rescaling
#107 opened by S-I-P - 1
Using on streaming video data
#104 opened by arjungoel7 - 1
dependencies issues
#121 opened by AnaRhisT94 - 1
Dependency with cuda 9.0
#81 opened by zeynepgokce - 1
customize actions class in the model
#112 opened by Tortoise17 - 0
missing videos
#118 opened by osamaibrahim13022 - 5
extract features
#100 opened by Serendipity-LC - 5
which one is the checkpoint?
#116 opened by tangyuelm - 4
Ideas to improve the training accuracy
#93 opened by Leslie-Fang - 3
How to train my i3d model
#98 opened by lixue17868800524 - 0
The problem of receptive field in I3D paper
#117 opened by ckx506772099 - 0
offline usage
#113 opened by Tortoise17 - 2
size of extracted features.
#106 opened by Joheun-Kang - 1
- 5
Video preprocessing code
#87 opened by javithe7 - 0
Training with different architectures
#111 opened by axelstram - 1
Train on videos of different length
#97 opened by kentwhf - 1
Has anyone tried to run this on mobile?
#78 opened by stev3 - 0
- 1
Inflating pre-trained models
#109 opened by varghesealex90 - 4
I3D on other dataset
#96 opened by kayleeliyx - 0
optical + Rgb
#105 opened by Joheun-Kang - 0
Pretrained weights on ImageNet only?
#101 opened by kentwhf - 0
flow_imagenet checkpoint
#102 opened by nik123 - 1
AttributeError: module 'tensorflow._api.v1.distribute' has no attribute 'experimental'
#91 opened by tempdeltavalue - 1
- 2
Difference between provided checkpoints.
#89 opened by QwCao - 0
- 2
Running on Composite Video file
#94 opened by harshsp31 - 2
Trouble saving I3d features
#86 opened by GayatriPurandharT - 0
Frame Numbers in Trained Model?
#90 opened by artest08 - 9
File format and pre-processing of RGB/Flow data
#79 opened by daveboat - 2
Successor network
#88 opened by ir0nt0ad - 2
Can you update the "requirements.txt"?
#84 opened by VijayRameshkumar - 0
Could I share your well-trained models on UCF-101 and HMDB-51 (RGB and Optical flow)
#83 opened by junmuzi - 0
training from scratch cost?
#82 opened by LindseyScharpman - 1
how to deal with long video and short video
#80 opened by zouying-sjtu - 2
How should I normalize the framerate?
#77 opened by skaws2003