Info about emo_detector_cli

[mastiffs92]

Hi , what is the image file path? I don't understand when I launch the script python for the training that passes as "image file path"? Thanks in advance for the answer!

[egottardo]

Hi, The "image file path" refers to the location where the original images for training is located.

If you are using Extended Cohn-Kanade database there are four folders:

1. The Images (extended-cohn-kanade-images.zip) - there are 593 sequences across 123 subjects which are FACS coded at the peak frame. All sequences are from the neutral face to the peak expression.
2. The Landmarks (Landmarks.zip) - All sequences are AAM tracked with 68points landmarks for each image.
3. The FACS coded files (FACS_labels.zip) - for each sequence (593) there is only 1 FACS file, which is the last frame (the peak frame). Each line of the file corresponds to a specific AU and then the intensity. An example is given below.
4. The Emotion coded files (Emotion_labels.zip) - ONLY 327 of the 593 sequences have emotion sequences.

Then, 'image file path' refers to the first (1) and 'emotion folder' refers to fouth (4).

Hope this helps. Ernani Gottardo

[mastiffs92]

I thank you for the answer, you mean this folder but I will pass this path does not work

[mastiffs92]

When to Use python script everything works in the testing phase, only that I don't see how files .cpp are recalled. I'm very interested for educational purposes.

[egottardo]

Excuse-me. I didn't observe that problem isn't in training phase (when creating new classification model). You are alread executing emo_detector_cli. In this case you should inform the path of one image. Below a figure of one test.

[mastiffs92]

Thanks for the answer, now I will try again !

[mastiffs92]

where am I wrong?

[egottardo]

It seems that the face was not locate by cbcl1 in the image. I'm also having trouble with cbcl1 in locating faces. I suggest that you test with another images or with haarcascade_frontalface_default.xml / haarcascade_frontalface_alt2.xml

[mastiffs92]

I have tested with your changes but situation isn't change.

[mastiffs92]

No idea? @egottardo

[dbellavista]

Try by adding some debug prints in facepreprocessor, to understand why it fails. Keep in mind also that you face has an hat. This can be a problem

On Mon, Jan 25, 2016 at 10:34 AM mastiffs92 [email protected] wrote:

No idea? @egottardo https://github.com/egottardo

— Reply to this email directly or view it on GitHub https://github.com/luca-m/emotime/issues/65#issuecomment-174449363.

[mastiffs92]

ok I will try , thanks.

[egottardo]

Its strange! For me it's working fine, but i'm using the model with 1vsallext. I suggest that you try with a model trained with 1vsallext.

[mastiffs92]

OK Thanks! I will try!

[r-sreesaran]

even i am getting the same error while using 1vsallext, Any ways too solve this problem

[mastiffs92]

@r-sreesaran Also with models 1_vs _all_ext you have same problems?

[r-sreesaran]

@mastiffs92 the error was happening because i was using .jpg file but now after running the below command, it throws up a error

./emo_detector_cli svm ../resources/haarcascade_frontalface_default.xml ../resources/haarcascade_eye.xml 52 52 1 5 8 ../compiled/dataset_svm_354_cbcl1_1vsall/classifiers/* Insert the image file path: /root/github/emotime/datasets/classifiers/svm/S032_002_01351426.png

OpenCV Error: Bad argument (The SVM should be trained first) in CvSVM::predict, file /root/essentials/opencv/opencv-2.4.9/modules/ml/src/svm.cpp, line 2131 terminate called after throwing an instance of 'cv::Exception' what(): /root/essentials/opencv/opencv-2.4.9/modules/ml/src/svm.cpp:2131: error: (-5) The SVM should be trained first in function CvSVM::predict

Note i am using the compiled version of binaries for linux

[mastiffs92]

I use the format of image file path '.png' and also i am using the compiled version of binaries for linux.

[sailsolar]

I am sorry but I am getting the following error. If you can kindly guide me out of it, that would be so awesome. What am I doing wrong? This is happening with both the master and the experimental version. I have also tried changing the parameters. I am using CK extended database files as the input image.

dell@gandalf:~/Desktop/emotime/emotime-1.2-experimental/assets$ ./emo_detector_cli svm ../resources/haarcascade_frontalface_default.xml ../resources/haarcascade_eye.xml 53 53 1 5 8 ../../svm_1vsallext_1_5_8_95c2eb0b58/*.xml
Insert the image file path: 
/home/dell/Downloads/cohn-kanade-images/S045/001/S045_001_00000001.png
Processing '/home/dell/Downloads/cohn-kanade-images/S045/001/S045_001_00000001.png'
OpenCV Error: Sizes of input arguments do not match (The sample size is different from what has been used for training) in cvPreparePredictData, file /home/dell/OpenCV/modules/ml/src/inner_functions.cpp, line 1114
terminate called after throwing an instance of 'cv::Exception'
  what():  /home/dell/OpenCV/modules/ml/src/inner_functions.cpp:1114: error: (-209) The sample size is different from what has been used for training in function cvPreparePredictData

Aborted (core dumped)

[egottardo]

In my tests, this error occured when the parameters WIDTH HEIGHT NWIDTHS NLAMBDAS NTHETAS (in your case 53 53 1 5 8) are different from the trainning phase.

[sailsolar]

@egottardo Yes you are right. I got it working after choosing the exact same parameter values for which the classifiers were trained on. Thank you for your support.

[rajat70595]

@egottardo @husnain-ahmad I tried with the same parameters as in the configuration file used to train the system. Following is the command I used :

./emo_detector_cli svm ../resources/haarcascade_frontalface_default.xml ../resources/haarcascade_eye.xml 32 32 1 5 6 /home/rajat/Downloads/emotime-master/dataset/classifiers/svm/*.xml

Insert the image file path: 
/home/rajat/Desktop/My_Project/google_004.jpg

But here is the error I am getting :

Processing '/home/rajat/Desktop/My_Project/google_004.jpg'
OpenCV Error: Assertion failed (dsize.area() || (inv_scale_x > 0 && inv_scale_y > 0)) in resize, file /build/buildd/opencv-2.4.8+dfsg1/modules/imgproc/src/imgwarp.cpp, line 1825
terminate called after throwing an instance of 'cv::Exception'
  what():  /build/buildd/opencv-2.4.8+dfsg1/modules/imgproc/src/imgwarp.cpp:1825: error: (-215) dsize.area() || (inv_scale_x > 0 && inv_scale_y > 0) in function resize

Aborted

It is working fine though with the CK+ database images in the png formats. Please help

[colceagus]

try scaling the google image(s) to the resolution CK+ images have :))

[egottardo]

I found the same problem with some small images. Since I did not have time to investigate the problem i temporarily manually increased the images.