innovarul / first-impressions Goto Github PK
View Code? Open in Web Editor NEWThe repository contains code, documentation of the code used in ChaLearn First Impressions Analysis challenge (ECCV - 2016)
License: Apache License 2.0
The repository contains code, documentation of the code used in ChaLearn First Impressions Analysis challenge (ECCV - 2016)
License: Apache License 2.0
Hi, it seems that I can't download the dataset from http://chalearnlap.cvc.uab.es/dataset/20/description/. I have to register to download the dataset, but there are some problems with the 'sign up' function. May you upload the dataset to google cloud or something else? Thank you!
Hi Arul,
Can we use the pre-trained models in your repository for feature extraction?
I installed OpenFace and pyAudioAnalysis libraries.
Can you please share the details to use this code for feature extraction alone without any training?
Links are not working for downloading training and test data set (train_val.txt). Any other links are available to download these data-sets ?
Hello,
I would like to test the pre-trained model LSTMModel#440.net and 3DCNNModel#600.net, which you uploaded.
However, I'm completely new to Lua, so would you release your testing code if you have the one?
Thank you!
Hi Arul,
Many thanks for your contribution. I wrote a script that executes a video through your trained LSTM/3DNN models. The 3DNN is working fine, but LSTM model throws the following errors. Any comment about my code would be greatly appreciated. Thanks in advance.
require 'torch'
require 'cutorch'
require 'cunn'
dofile 'utilities.lua'
require 'rnn'
require 'nngraph'
logger = dofile 'log.lua'
VIDEOFEATURESROWS = 6
VIDEOFEATURESCOLS = 0
VIDEOFEATURESLMROWS = 6
VIDEOFEATURESLMCOLS = 0
opt = {}
opt.targetScaleFactor = 1
opt.type= 'float'
--opt.type= 'cuda'
opt.LSTM = true
mp4name = 'somename.mp4'
--audio data
validationAudioFeaturePath = 'data/validationaudiofeat';
validationaudiofiles = dir.getallfiles(validationAudioFeaturePath)
validationAudioData = loadAudioFeaturesFromFolder(validationaudiofiles)
local inputaudio = validationAudioData[mp4name]
if opt.type == 'float' then
inputaudio = inputaudio:float();
else
inputaudio = inputaudio:cuda();
end
--video data
local validationVideoData = prepareVideoFramesData({[mp4name]=1}, VIDEOFEATURESROWS, 'validation',opt.LSTM)
if(opt.type == 'cuda') then
validationVideoData[mp4name]= validationVideoData[mp4name]:cuda()
else
validationVideoData[mp4name] = validationVideoData[mp4name]:float()
end
--all post-processed inputs
local currentInput = {inputaudio, validationVideoData[mp4name] }
--load the traied model
if(not opt.LSTM) then
model = torch.load('models/3DCNNModel#600.net')
else
model = torch.load('models/LSTMModel#440.net')
end
model:evaluate()
local pred = model:forward(currentInput)
--building a tensor to add the results
validationPredictions = {}
validationPredictions = torch.Tensor(5):fill(0)
if(not opt.LSTM) then
validationPredictions = validationPredictions + torch.squeeze( pred:double() / opt.targetScaleFactor )
else
validationPredictions = validationPredictions + torch.squeeze( (torch.mean(pred:double(),1) / opt.targetScaleFactor) )
end
The errors:
/root/torch/install/bin/luajit: /root/torch/install/share/lua/5.1/nn/Container.lua:67:
In 5 module of nn.Sequential:
/root/torch/install/share/lua/5.1/rnn/LSTM.lua:188: attempt to perform arithmetic on field 'step' (a nil value)
stack traceback:
/root/torch/install/share/lua/5.1/rnn/LSTM.lua:188: in function </root/torch/install/share/lua/5.1/rnn/LSTM.lua:187>
[C]: in function 'xpcall'
/root/torch/install/share/lua/5.1/nn/Container.lua:63: in function 'rethrowErrors'
/root/torch/install/share/lua/5.1/nn/Sequential.lua:44: in function 'forward'
5_validate.lua:70: in main chunk
[C]: in function 'dofile'
/root/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:150: in main chunk
[C]: at 0x00405d50
WARNING: If you see a stack trace below, it doesn't point to the place where this error occurred. Please use only the one above.
stack traceback:
[C]: in function 'error'
/root/torch/install/share/lua/5.1/nn/Container.lua:67: in function 'rethrowErrors'
/root/torch/install/share/lua/5.1/nn/Sequential.lua:44: in function 'forward'
5_validate.lua:70: in main chunk
[C]: in function 'dofile'
/root/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:150: in main chunk
[C]: at 0x00405d50
Hello,
I am able to import the file
by
require "nn"
require "cunn"
require "rnn"
net = torch.load("LSTMModel#440.net")
but I don't understand how to use it.
it throws an error saying
torch.load("LSTMModel#440.net")
/home/conankapoor/torch/install/bin/luajit: ...or/torch/install/share/lua/5.1/rnn/AbstractRecurrent.lua:287: attempt to call field 'tostring' (a nil value)
stack traceback:
...or/torch/install/share/lua/5.1/rnn/AbstractRecurrent.lua:287: in function <...or/torch/install/share/lua/5.1/rnn/AbstractRecurrent.lua:283>
[C]: in function 'tostring'
...onankapoor/torch/install/share/lua/5.1/nn/Sequential.lua:118: in function <...onankapoor/torch/install/share/lua/5.1/nn/Sequential.lua:107>
[C]: in function 'tostring'
...e/conankapoor/torch/install/share/lua/5.1/trepl/init.lua:262: in function 'rawprint'
...e/conankapoor/torch/install/share/lua/5.1/trepl/init.lua:302: in function 'print'
...e/conankapoor/torch/install/share/lua/5.1/trepl/init.lua:663: in function 'repl'
...poor/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:204: in main chunk
[C]: at 0x00405d50
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.