You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently the cars are broadly classified into 3 angles, -45 degrees, +45 degrees, 0 degrees. It is a classification task and not regression as it should be.
A network for regression needs to be developed that can take an image as input and label the angle from anywhere between -45 to 45 degrees.
All the literature so far does not tell us a nice way to do this. The links related to this issue are as follows:
Currently the cars are broadly classified into 3 angles, -45 degrees, +45 degrees, 0 degrees. It is a classification task and not regression as it should be.
A network for regression needs to be developed that can take an image as input and label the angle from anywhere between -45 to 45 degrees.
All the literature so far does not tell us a nice way to do this. The links related to this issue are as follows:
BVLC/caffe#512
BVLC/caffe#881
BVLC/caffe#1765
https://groups.google.com/forum/#!topic/caffe-users/o4cpDNylo3Q
https://groups.google.com/forum/#!topic/caffe-users/8iE_t-P_m3M
http://corpocrat.com/2015/02/24/facial-keypoints-extraction-using-deep-learning-with-caffe/
@satwikkottur will be using this.
The text was updated successfully, but these errors were encountered: