FloodNet Challenge @ EARTHVISION 2021 - Track 1 Forum

Go back to competition Back to thread list Post in this thread

> IOError submission

Hi, thanks for this interesting challenge.

When i try to submit a prediction i get the following trace stack, which doesn't really help me figuring things out:
Traceback (most recent call last):
File "/tmp/codalab/tmpoIyDl5/run/program/evaluate.py", line 122, in
pred = png.read_png_int(predict_path)
IOError: [Errno 2] No such file or directory: '/tmp/codalab/tmpoIyDl5/run/input/res/10822.png'

I submit a zip file, which contains masks for all images with the name xxxx.png and one file "image_classes.json"

I don't know why the 10822.png file will not exist, I guess the 10822.jpg RGB images' size is 4592*3072,when I submit the predicet mask,I resize the mask whose size is 4592*3072 to 4000*3000,
but it still reports the wrong result.

I guess your testing procedure is based on the documented order,before predicting 10822.png, error will not occurred, such as when predicting 10816,10815.png.

Thanks!

Posted by: WIDEA @ April 7, 2021, 2:55 a.m.

Hello,

Thank you for your submission. I checked your zip file. When I unzip it, it creates another layer of folder. For example, if the program is looking for "1010.png" file and the submitted file is "submission.zip", then when unzipping the folder it should have the following structure: "submission/1010.png", instead after unzipping your folder structure is as follows: "submission/submission/1010.png". An extra layer is created after unzipping. As a result the evaluation script could not find the files. Please make sure that during zipping the masks and json file you are zipping the files, not the folder. Hope this helps.

Thank you!

Posted by: binalab @ April 7, 2021, 3:25 a.m.

Also you don't have to resize the predicted masks during the submission. You can keep it as the same size as original images. The evaluation script will resize them before the calculating the mIoU.

Thank you!

Posted by: binalab @ April 7, 2021, 3:30 a.m.

I am getting this error
WARNING: Your kernel does not support swap limit capabilities or the cgroup is not mounted. Memory limited without swap.
Traceback (most recent call last):
File "/tmp/codalab/tmp8Z_aZP/run/program/evaluate.py", line 122, in
pred = png.read_png_int(predict_path)
TypeError: Object does not appear to be a 8-bit string path or a Python file-like object

kindly suggest

Posted by: avinash_nesac @ April 22, 2021, 4:34 a.m.

@avinash_nesac, please make sure that the submitted folder does not have any extra layer folders. Also the content of the predicted masks is 8 bit. You can check the data type of masks of the train images.

Thank you!

Posted by: binalab @ May 11, 2021, 8:22 p.m.
Post in this thread