Open darthgera123 opened 2 years ago
It seems the issue is with the multithreaded approach of reading the files from the dataloader. The error is coming when it is trying to load the images. Is there a version which is not multithreaded?
@darthgera123 Thank you for taking an interest in our work. I'm sorry for that we didn't have a non-multithreaded version, but you can try to set the loading threads number from 4 to 1 in src/model.py: 151.
@LWT3437 same issue still coming. Tried with different tensorflow versions as well :(
@darthgera123 Oh no, I guess the data processing you did doesn't match the training code. However, I don't have the conditions for debugging anymore, so I am afraid it is difficult to help you solve this error QAQ
Actually we didn't provide a guide on how to retrain and don't recommend using this code for retraining since it was written a long time ago and is now outdated. The core ideas of our paper is calibration preprocessing and scale-invariant loss, I suggest re-implementing based on these two ideas, if you want to do a further work.
@darthgera123 Hi, I also encountered this problem, have you solved it?
@darthgera123 Hi, I also encountered this problem, have you solved it?
Not really. I just reimplemented a version here
Hi Wonderful work. While running the code for training im getting this error:
I am unable to figure this out, please help @LWT3437