hszhao / ICNet

ICNet for Real-Time Semantic Segmentation on High-Resolution Images, ECCV2018

Home Page:https://hszhao.github.io/projects/icnet

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

shape of input image in cityscapes dataset

coderzbx opened this issue · comments

why the input image size of cityscapes is 1025 x 2049?
Shouldn't be 1024 x 2048? The original image size is 1024 x 2048 in cityscapes dataset.

I have also same issue. What is the input image size?

@gangooteli Does your Cityscapes test dataset have labels?I have downloaded a label without a test data set from the official website.So I can't get the test mean IOU. Thank you!

@InstantWindy test set's label is withheld by CityScapes test server, you need to submit to evaluate on test set.

same question as @coderzbx

@baoruxiao Should I submit the tested picture and get the result?What should I submit to this website? Is the predicted tag value or the predicted image? Thank you!

you need to submit mask and txt, following the instruction on cityscapesscript repo

@baoruxiao I'am a novice.Only this simple indication on the cityscapes website,
image
I still don't quite understand how to upload after reading the instructions.Where is the cityscapesscript repo you are talking about?Can you send me ?Thank you !

@coderzbx

why the input image size of cityscapes is 1025 x 2049?
Shouldn't be 1024 x 2048? The original image size is 1024 x 2048 in cityscapes dataset.

image
Hope this is applicable here also.
Refer: https://github.com/tensorflow/models/blob/master/research/deeplab/g3doc/faq.md