Closed BoopSainy closed 2 years ago
Thank you for your interest. After the stage-1 is finished, it will run the function named as scan to generate the attention map for all the aerial-view images. Can you check the log of scan and see how many batches are there? It seems that some images are not scaned which is weird. Did you make any other changes?
Hi,
Thanks for your reply! The log of scan is lost, let me restart the training from the last epoch and see the batches number.
I only adjusted the epoch from 50 to 10, and adjusted the "localhost" in dist_url.
An update, the number of attention maps in "./result_vigor/attention/train/" is 46284 not 50608
Could I know what is the meaning of
"self.delta_unit = [0.0003280724526376747, 0.00043301140280175833]"
in the VIGOR class?
Really thanks for your help!
I see. The problem is the sampling. I just updated the code and you can pull and try again.
The delta unit is the GPS offset of half an image. But it not used in this codebase, you can ignore it.
Could I know what is the meaning of
"self.delta_unit = [0.0003280724526376747, 0.00043301140280175833]"
in the VIGOR class?
Really thanks for your help!
Sure. I will try it now. Really thanks for your help and answering.
There is still the same issue when running the latest code. The error is:
FileNotFoundError: [Errno 2] No such file or directory: './result_vigor/attention/train/12539.png' (1415.png last time)
There are 46199 attention maps in "./result_vigor/attention/train" now.
The batches of scan are 5001 and 5664, respectively.
I made a mistake in my last commit. Hopefully, the new version will work. :)
There is still the same issue when running the latest code. The error is:
FileNotFoundError: [Errno 2] No such file or directory: './result_vigor/attention/train/12539.png' (1415.png last time)
There are 46199 attention maps in "./result_vigor/attention/train" now.
The batches of scan are 5001 and 5664, respectively.
Really thanks for your help at such a late night!
I will try it now and continue to deeply understand transgeo codebase.
Have a good night! :)
Hi Siji,
I am still going through your code. But I find the changing number of attention maps in "attention/train" file might be caused by line 339, 340 in "train.py".
If we use "train_dataset" but not "train_scan_dataset" in "train_scan_loader", not all positive panoramas will be loaded and the indexes of attention maps here depends on the number of positive panoramas.
I will test and confirm it.
Songyang
Thank you so much for testing it ! I think replacing train_dataset with train_scan_dataset should work. I'll update the code when it is confirmed.
Hi Sijie,
I've tested it and it works now. Just need to change "train_dataset" to "train_scan_dataset" of "train_scan_loader" in train.py
Have a good night!
Thank you so much for testing it. I have updated the code and it should be good now.
I'll close the issue if there is no other question.
I'll close the issue if there is no other question.
Sure! Thanks for your sharing!
hello, thanks for your sharing project. I meet the same question with BoopSainy when I try to train model on VIGOR. I changed the code as you modified last month, but it still unable to work. Can you tell me what should I do to fix it? Thanks again!
FileNotFoundError: [Errno 2] No such file or directory: './result_vigor/attention/train/35112.png'
Hi,
Really thanks for your open-source code and VIGOR dataset!
I meet a strange issue when I try to learn your code. I run the "run_VIGOR.sh" file and adjusted the epoch to 10 for both stage. When the stage-1 completed and automatically skipped to stage-2. There is an error:
FileNotFoundError: [Errno 2] No such file or directory: './result_vigor/attention/train/1415.png'
Then I check the /attention/train folder and found there is only 50608 png attention images without the 1415.png Do you once meet similar issue when running the code?