iitkliv / dlvcnptel Goto Github PK

Hi,

Code is old and having many issues in Pytorch 1
Please upgrade your code

Sir, this code is not running on cpu version of pytorch.and asking for cuda

Hi, I am facing issue while running this jupyter notebook (Lec 14- ALL-IDB dataset) on Python 3 after fixing the code. During the validation of the classifier it is giving the testing accuracy as 0.0 for all epochs. However the code runs fine for Python 2.
I really cannot get the issue. Please Help..!

Hi,

I got this error, when running the lab lecture 3 on the line

plt.imshow(lbp_img,cmap='gray') #Displaying LBP

It would be really great if you suggested a workaround.

matplotlib and scikit-image versions i used...

matplotlib: 2.0.2-np113py27_0
scikit-image: 0.13.1-py27h14c3975_1

Hey, @rachana-sathish ,
Gabor filter gives output a complex number, with coefficients of real and imaginary part as follows
gaborFilt_real,gaborFilt_imag = gabor(img_arr,frequency=0.6)
For feature extraction, we are interested in magnitude of the output, therefore, we do the following
gaborFilt = (gaborFilt_real**2+gaborFilt_imag**2)//2

However, shouldn't be np.sqrt((gaborFilt_real**2+gaborFilt_imag**2)), |z|=sqrt(r^2+i^2)

I guess, it won't have any effect on training of the model, as the real and imaginary part have been incorporated in some form.

Thanks for your time and effort.

Code in the repo is old. Please update it with the video/course lectures.