1
$\begingroup$

little help here

    "I don't want to put unknown encodings of group image in to a classifier"

    what i did was find Matches for each of the encoding and trying to remove   encodings with matches=0

for encoding in encodings:

  matches=np.count_nonzero(face_recognition.compare_faces(data["encodings"],
            encoding))         
        d["encods"]=[matches,encoding]
    for k,v in list(d.items()):
        if v[0]==0:
        del d[k]

        print("dict",d)

I am not getting what i want.Please help

$\endgroup$
  • $\begingroup$ Which algorithm is it? Is it a lookup table search in a dictionary? $\endgroup$ – Manuel Rodriguez Oct 15 '18 at 6:27
  • $\begingroup$ I am using Random Forest classifier $\endgroup$ – Akhil Alexander Oct 15 '18 at 6:34
  • $\begingroup$ As far as i understand, you have a database with images who are all faces, and use the Random Forest classifier of OpenCV for what? Matching means, to take picture1 and search in the database for it. Is that the task, to identify a picture in a database of other pictures? $\endgroup$ – Manuel Rodriguez Oct 15 '18 at 6:54
  • $\begingroup$ I train my database of images and find out the facial encodings using opencv's facial encodings then store it in to a pickle file ,later i split these encodings and label in to test and train and trian it using Random forest(it gives more accuracy) .whenever a picture(both known and unknown) is given ,will find out facial encodings using opencv and compare these encoding with stored encodings if these found a match i will put it in to the classifier for prediction or else it will print unknown that was my earlier plan $\endgroup$ – Akhil Alexander Oct 15 '18 at 7:08
  • $\begingroup$ but a group of image with unknown encoding will also go through classifier and make prediction ,so I thought i could remove the unknown encodings and rest of can be passed through the classifier $\endgroup$ – Akhil Alexander Oct 15 '18 at 7:09
1
$\begingroup$

In face recognition it is very important to choose the right image encoding. The good one are VLAG, Fisher and Bag of Words. The “The ORL Database of Faces” offers more encodings. If the correct encoding was chosen it become much easier to train the neural network.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.