what is the real of output in kitti_eval.py

[ywangeq]

in evaluation fuction ，we can get the output from sess.run, can anyone tell its real meaning? I mean that every elements in the matrix, because if I use output[0][:,k] to de the multiple class , how can I correspond to the original color? The element in the output is the confidence of class (or the index of the class with max confidence?

[obendidi]

output[0] is a tensor with shape [xxxx , NUM_CLASSES] , as it is the output of a softmax layer (confidence of each class) , you could simply add an argmax function to get the index of the class with biggest confidence and finally get a tensor with shape [xxx] that you then can reshape to the same shape as your image array [height , width ] PS : you should note that the image that you get is gray scale and all pixels are <= NUM_Classes , so to visualize it you need to map each of those classes to and RGB color for example

[ywangeq]

@bendidi ,Thanks for your response, i will check it again. It is not very convenient to check the output of image by print every loop. So i also have some mistakes here and get a image with all orange pixels back! hhhhh, thanks you very very much, i think i can have some ways to debug it !!!!!!