You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In your paper, Table 10 shows that models like FCN, Unet, and DANet did not achieve the results as presented in the table, even after training and testing them twice. For example, in the case of DANet, I trained it using a single NVIDIA 3090 GPU. I made some adjustments such as modifying your multi-GPU setup to a single GPU and setting the batch size to 8 (both 4 and 8 gave similar results). Everything else remained the same. Can you provide any suggestions? Does using a single GPU affect the accuracy, or are there any other crucial parameter settings that I might have overlooked? Please advise. Thank you.
The text was updated successfully, but these errors were encountered:
In your paper, Table 10 shows that models like FCN, Unet, and DANet did not achieve the results as presented in the table, even after training and testing them twice. For example, in the case of DANet, I trained it using a single NVIDIA 3090 GPU. I made some adjustments such as modifying your multi-GPU setup to a single GPU and setting the batch size to 8 (both 4 and 8 gave similar results). Everything else remained the same. Can you provide any suggestions? Does using a single GPU affect the accuracy, or are there any other crucial parameter settings that I might have overlooked? Please advise. Thank you.
The text was updated successfully, but these errors were encountered: