Marek

Hi, 1) The face detection bounding boxes are provided in the training set (300-W dataset). 2) I don't really understand this question, could you elaborate? Best regards, Marek

Hi, Thanks for the clarification, here are my answers: 1) I do not know what face detector the authors of the dataset used. All they specify is that "The face...

Hi, The value os self.initLandmarks never changes, it is always the mean face shape located in the middle of a 112x112 image. Marek

Hi, Do you mean that the second stage no longer gets a face image as input? If that's the case then it probably won't work. You can provide the second...

Hi, It sounds like the datasets were not generated correctly, division by 0 suggests that there are no images in the dataset. Did you download the images from the ibug...

Have you tried debugging this? Does a key like that exist in the dictionary? Do the file extensions match? Marek

Hey, This looks like a theano issue to me. Do you use the version I specified in the readme? Have you tried searching theano-related GitHub issues? Also, what is in...

Hi, Did you try running some other theano examples to determine whether the issue is in the DAN code (this repo) or with your Theano installation in general? Thanks, Marek

Hi, I usually do not train for a specific number of epochs, like you specified here but rather till the valdation error stops decreasing rapidly. As for the rest: it...

Hi, I don't know the specific number of epochs, I would train the first stage until the validation stops improving at a reasonable rate (like a new best error value...