cleardusk / emotionchallenge Goto Github PK
View Code? Open in Web Editor NEWSource code for 1st winner of face micro-emotion competition, FG 2017.
Source code for 1st winner of face micro-emotion competition, FG 2017.
So, I have a couple of questions and I really appreciate if you could help:
Do you have pretrained model in which I could simply feed in an image and get the emotion out of your system? if that is the case where should I feed in the image?
How are you doing the training process? I don't get where is the past to the training dataset? It is not mentioned in your Read.me neither it is mentioned which dataset you are using for training. Can you please share those information?
Thanks a lot,
Mona
Here are all the steps I took but really don't make any sense because I don't even know how to feed in a test image or how to set the path to training images
[jalal@goku cs585]$ cd EmotionChallenge/
[jalal@goku EmotionChallenge]$ ls
cnn crop_align data fact_sheets.pdf fact_sheets.tex models readme.md
[jalal@goku data]$ file train_data
train_data: broken symbolic link to `/data/gjz/Training'
[jalal@goku data]$ cd ..
[jalal@goku EmotionChallenge]$ cd crop_align/
[jalal@goku crop_align]$ mkdir build
[jalal@goku crop_align]$ cd build/
[jalal@goku build]$ cmake ..
-- The C compiler identification is GNU 4.8.5
-- The CXX compiler identification is GNU 4.8.5
-- Check for working C compiler: /usr/bin/cc
-- Check for working C compiler: /usr/bin/cc -- works
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Detecting C compile features
-- Detecting C compile features - done
-- Check for working CXX compiler: /usr/bin/c++
-- Check for working CXX compiler: /usr/bin/c++ -- works
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- OpenCV: /scratch/sjn-p2/anaconda/anaconda2/include/opencv/scratch/sjn-p2/anaconda/anaconda2/include
-- Configuring done
-- Generating done
-- Build files have been written to: /scratch/mona/download/cs585/EmotionChallenge/crop_align/build
[jalal@goku build]$ make
Scanning dependencies of target crop_align
[ 33%] Building CXX object CMakeFiles/crop_align.dir/crop_align.cpp.o
[ 66%] Building CXX object CMakeFiles/crop_align.dir/imtransform.cpp.o
[100%] Linking CXX executable crop_align
[100%] Built target crop_align
[jalal@goku build]$ cd ..
[jalal@goku crop_align]$ ls
build CMakeLists.txt crop_align.cpp crop_align.py imtransform.cpp imtransform.h __init__.py landmark.py util.py
[jalal@goku crop_align]$ cd ..
[jalal@goku EmotionChallenge]$ cd cnn
[jalal@goku cnn]$ ls
data_layer.py deploy.prototxt extract.py landmark_224.py net.prototxt prepare_data.py solver.prototxt train_val.sh util.py
[jalal@goku cnn]$ python prepare_data.py
[jalal@goku cnn]$ python train_val.sh
File "train_val.sh", line 3
caffe train -solver solver.prototxt -gpu 3 2>&1 | tee final_03051346.log
^
SyntaxError: invalid syntax
Additionally, I am not sure what the line of code that is causing error is trying to do and what is the fix to it.
Additionally, after I run
1168 cd cnn/
1169 ./prepare_data.py
1192 ./train_val.sh
I get this error:
[jalal@goku cnn]$ ./train_val.sh caffe: symbol lookup error: caffe: undefined symbol: _ZN6google14FlagRegistererC1EPKcS2_S2_S2_PvS3_
Do you how should this be fixed?
Using Python2.7 I get this error:
(py2emotion) [jalal@goku crop_align]$ /scratch/sjn-p2/anaconda/anaconda2/bin/python landmark.py
Traceback (most recent call last):
File "landmark.py", line 7, in <module>
import dlib
File "/scratch/sjn-p2/anaconda/anaconda2/lib/python2.7/site-packages/dlib/__init__.py", line 1, in <module>
from .dlib import *
ImportError: libcudnn.so.6: cannot open shared object file: No such file or directory
(py2emotion) [jalal@goku crop_align]$ which conda
/scratch/anaconda3/envs/py2emotion/bin/conda
(py2emotion) [jalal@goku crop_align]$ source deactivate py2emotion
[jalal@goku crop_align]$ /scratch/sjn-p2/anaconda/anaconda2/bin/python landmark.py
Traceback (most recent call last):
File "landmark.py", line 7, in <module>
import dlib
File "/scratch/sjn-p2/anaconda/anaconda2/lib/python2.7/site-packages/dlib/__init__.py", line 1, in <module>
from .dlib import *
ImportError: libcudnn.so.6: cannot open shared object file: No such file or directory
[jalal@goku crop_align]$ /scratch/sjn-p2/anaconda/anaconda2/bin/conda install cudnn=6.0
Fetching package metadata .............
Solving package specifications: .
UnsatisfiableError: The following specifications were found to be in conflict:
- caffe-gpu -> cudnn ==5.1
- cudnn 6.0*
Use "conda info <package>" to see the dependencies for each package.
[jalal@goku crop_align]$
Using Python 3, I get this other error. No error on python landmark.py but in the command following it:
[jalal@goku crop_align]$ python landmark.py
[jalal@goku crop_align]$ cd ..
[jalal@goku EmotionChallenge]$ ls
cnn crop_align data fact_sheets.pdf fact_sheets.tex models readme.md
[jalal@goku EmotionChallenge]$ cd cnn/
[jalal@goku cnn]$ ls
data_layer.py deploy.prototxt extract.py landmark_224.py net.prototxt prepare_data.py __pycache__ solver.prototxt train_val.sh util.py
[jalal@goku cnn]$ python extract.py
Traceback (most recent call last):
File "extract.py", line 78, in <module>
submit()
File "extract.py", line 74, in submit
extract('../models/final.caffemodel')
File "extract.py", line 58, in extract
assert len(landmark) == 136
TypeError: object of type 'map' has no len()
If you could help to solve preferrably the one that gives an error for Python 3 and if not the one for Python 2.7. Thanks.
我是FG2020复合表情的参赛者,请问你有上一届比赛的论文模板吗?另外最终测试集的解压密码也没有公布,想请问一下上一届的比赛最终测试阶段是什么流程?
So, I get this error for Python 3:
[jalal@goku cnn]$ python extract.py
Traceback (most recent call last):
File "extract.py", line 78, in <module>
▽
submit()
File "extract.py", line 74, in submit
extract('../models/final.caffemodel')
File "extract.py", line 61, in extract
pred = classifier.predict(img, landmark)
File "extract.py", line 32, in predict
self.net.blobs['data'].data[0, ...] = image_array.transpose((2, 0, 1))
▽
AttributeError: 'NoneType' object has no attribute 'transpose'
Looking at the code:
def extract(model_file):
classifier = Classifier(model_file=model_file,
deploy_file='deploy.prototxt')
records = open('../data/test_ld.txt').read().strip().split('\n')
# print(records[:1])
predicts = []
for rec in records:
rec = rec.split()
fp = rec[0]
# print(fp)
landmark = rec[1:]
landmark = map(float, landmark)
# print(landmark)
assert len(list(landmark)) == 136
It is looking at this file:
../data/test_ld.txt
Looking at this file:
I have something like
../data/face_224/104/Sample_104_001.JPG
I expect to see a folder named 104 in the face_224 folder. However, there is no folder here and only a readme.md file which says 224x224 size image stores here.
[jalal@goku data]$ cd face_224/
[jalal@goku face_224]$ ls
readme.md
So I am confused what has gone wrong and how to fix it?
Any suggestion is really appreciated. I had no problem with the first step which was done in the crop_align folder:
[jalal@goku crop_align]$ python landmark.py
[jalal@goku crop_align]$
From what I understand the crop didn't work because you are stating that you expect the cropped folder to be in data/face_224 folder in which face_224 is the name of my image however, there's nothing in that folder.
Any help is really appreciated.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 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. 📊📈🎉
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 ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.