Giters

hq-jiang / instance-segmentation-with-discriminative-loss-tensorflow

Tensorflow implementation of "Semantic Instance Segmentation with a Discriminative Loss Function"

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

how to set the num_classes?

ygren opened this issue · comments

commented

hello, than you for your wonderful work! I try your work to my dataset, but there is a question. my picture is 5~18 lane line per image ,so I draw my lane line in 19 different colors. and you use 6 different colors，the num_classer is 12. but when I change my num_classes like you in 38. There are some mistakes. but when i change it to 12.Training can be done successfully . So how do I set up this parameter？
Can you give me some advice？
thx

commented

Hi, thank you! The num_classes parameter refers to the original ENet and has to be fixed at 12, so it can be loaded from the checkpoint file. This layer is eventually replaced, anyway. The number of instances is not something that is represented as a parameter in this concept. If you have more instances you could also think about increasing the feature_dim. You would have to adapt the visualization for this.