Tests of Emotions Detection from a Video

Share it with your friends Like

Thanks! Share it with your friends!


Tests of the Emotion Detector created in Python. As it can be seen, the detector prioritizes neutral and happiness expressions (this was due to the datasets employed in the creation of the detector; they had many more samples of these expressions than the others).

The complete source code (under MIT license) is available in Github: https://github.com/luigivieira/emotions

The original video used in the tests is the Human Emotions, used and reproduced here with the kind authorization from the folks of the Imagine Pictures video production company (thank you guys!). Original URL of the video: https://www.youtube.com/watch?v=UjZzdo2-LKE


SPC says:

How can I calculate and display those values for particular expression, pls help in code

M. Revaldi Anggara says:

What method are you using?

TeaJay says:

Any way to make this more stable? The values are changing a lot although the face impressions don't change.

Ainfluenced says:

Have you tried your project in other datasets except these datasets? I have a work that I classify with cnn by creating geometrical shapes from facial landmark points. Although the results are very good in the ck dataset, it is not good in other datasets. Jaffe, Toronto, Kdef (frontal) are some of these data sets.

Nilesh Aggarwal says:

Hey can someone please help me with the data set and the technology involved in this algorithm?

ali hosseinzade says:

tnx a lot sir

Jhonatan Souza says:

Olá Luiz, vi que é brasileiro. Você que implementou esse algoritmo ou apenas rodou ? tem o artigo com mais informações sobre ele ? método de treinamento, e etcs ?

Tim J.P says:

Hey Luiz, I'd love to use a short extract of your video (< 3 sec) for a trailer for a university project. It's a video about data ownership in the age of AI. The project is not commercial in any manner. Are you okay with that? Can you grant me a permission for that? Of course I would give you credit! Thank you in advance

alemayehu Wakjira says:

Is it possible to find a training/test dataset for happy, angry and neutral facial expressions?

Luis Lerma says:

hello, I have read everything related to this code here and in [github.com], but I do not give a guide that leads me to implement the code in a correct way.

I need information on how to solve it, maybe I want to use it in part of my thesis, could you help me run the code? I already have all the libraries and others but I get these errors.

thank you very much

C:Python27libsite-packagesskimagefilters_gabor.py:76: RuntimeWarning: divide by zero encountered in double_scalars
sigma_x = _sigma_prefactor(bandwidth) / frequency
C:Python27libsite-packagesskimagefilters_gabor.py:78: RuntimeWarning: divide by zero encountered in double_scalars
sigma_y = _sigma_prefactor(bandwidth) / frequency
C:Python27libsite-packagesskimagefilters_gabor.py:81: RuntimeWarning: invalid value encountered in double_scalars
np.abs(n_stds * sigma_y * np.sin(theta)), 1))
C:Python27libsite-packagesskimagefilters_gabor.py:83: RuntimeWarning: invalid value encountered in double_scalars
np.abs(n_stds * sigma_x * np.sin(theta)), 1))
Traceback (most recent call last):
File "C:/Users/Luiis Lerma/Desktop/Software Tesis/emotions-master/test.py", line 503, in <module>
File "C:/Users/Luiis Lerma/Desktop/Software Tesis/emotions-master/test.py", line 244, in main
data = VideoData()
File "C:/Users/Luiis Lerma/Desktop/Software Tesis/emotions-master/test.py", line 57, in _init_
self._bank = GaborBank()
File "C:UsersLuiis LermaDesktopSoftware Tesisemotions-mastergabor.py", line 153, in _init_
kernel = gabor_kernel(frequency, orientation)
File "C:Python27libsite-packagesskimagefilters_gabor.py", line 84, in gabor_kernel
y, x = np.mgrid[-y0:y0 + 1, -x0:x0 + 1]
File "C:Python27libsite-packagesnumpylibindex_tricks.py", line 178, in _getitem_
int(math.ceil((key[k].stop – start)/(step*1.0))))
OverflowError: cannot convert float infinity to integer

ibtissam chouho says:

i hve a database 10k and coude source but the programe is not succefls

Ke Ke says:

Can I get a source of your test video? How did you train the classifier?

over00lord Unknown says:

This thing seems to think that they are neutral most of the time… Odd… But I guess that is to be expected with anything new and experimental. Keep up the good work! 🙂

Drukan says:

Olá, tudo bem? Eu achei seu vídeo em uma dúvida no Stack overflow, eu estou tentando verificar se em uma foto a pessoa está neutra, saberia o melhor jeito de fazer isso ? Eu estou Opencv pra capturar o rosto e os olhos, entretanto o haarcascade do sorriso que já vem no Opencv não funciona de maneira eficaz. Estou fazendo em Java

Big Mouth says:

I'm in Start but stuck with data set not able to find it

Big Mouth says:

Hello sir , I'm also trying to learn emotion detection will you please guide me ?

Write a comment


Area 51