2014-10-02 03:59:42 -0600 | commented question | Face recognition, different results for same image @chebhou: I do plan to align the faces, but right now I tested with the exact same face, the alignment is exactly the same, but the face is cropped differently... |
2014-10-02 03:28:45 -0600 | commented question | Face recognition, different results for same image no, I don't use javacv, I built the jar from opencv 2.4.9 and for FaceRecognizer I used a tutorial written by you I think |
2014-10-02 01:46:40 -0600 | commented question | Face recognition, different results for same image thanks for that, I was really not sure. I think I saw somewhere that the confidence was inverse proportional to the distance and I got confused. Also, when I tested with the face database unprocessed, I got values in the 80-90 range and I thought they were a point of reference. Could you please tell me another thing. Is the confidence the smallest value from the trained images? Or is it an average of some sort? |
2014-10-01 16:38:41 -0600 | asked a question | Face recognition, different results for same image Hello, I am working on a face recognition project. These are the pre-processing steps I use:
After this I use the images resulted to train a LBPH face recognizer. From my tests it seems that the recognizer returns a confidence that represents the best match between the tested image and another image in the training set(please correct me if I'm wrong). I tried training the model with 15 images and then trained it with only the best matching image and I got the same exact confidence in both cases. To verify the accuracy I tried this: I took an image, used it for training. Cropped the image manually in gimp(the face is not affected by the crop). And I tested the original image against the cropped image and only got a confidence of 42.48...(the range seems to be between 0..100, identical images giving 0.0 as a result, so I'm not sure if it's good to have a big confidence or not, either way 42 is in the middle so the confidence is not very good). I argue that the confidence should be very good since the faces in the images are identical. I observe 2 problems here:
I use the default values for the "train", "predict" and "detectMultiScale" methods. Maybe this is something that I could improve, but I'm not sure how. Note. This is probably not important, but I use java with opencv on an Ubuntu machine. |