Model Flops

[liux520]

It's a remarkable work. However, I had some problems in the reproduction. The paper reports 0.62M counts and FLOPS:9.7G at 1024x512 scale, but the model code provided for testing has 0.61M counts and FLOPS:15.941G. Can you check the model file? I am looking forward to your reply!

[liux520]

The settings of the paper (Blocks number : [3,3,20,3,3]) has 0.67M counts and FLOPS:16.09G. (1024x512 scale)

[liux520]

Do you solve the problem? If you really can't find the issue, you can follow your current path to modify the code. By the way, I found that two 2080ti can't run successfully (the paper says single 2080ti)? We can add a friend to discuss together (are you Chinese?), I am also a beginner who just got into semantic segmentation not long ago.

[qqwewewejekl]

It's a remarkable work. However, I had some problems in the reproduction. When I execute this code, I keep getting an error。The code is" with open(self.data_dir + '/' + fileName, 'r') as textFile:"I would like to ask what causes this. I am looking forward to your reply.Thank you.

hi,did u solve this problem,i think the dataset path was wrong.