search

tyb311 / TCCT

MICCAI2022 GOALS Challenge & Paper accepted by TMI2023 (Retinal Layer Segmentation in OCT images with Boundary Regression and Feature Polarization)
https://ieeexplore.ieee.org/document/10255666
13 stars 0 forks source link

pretrained model setting #2

Open ChenShunHsu opened 11 months ago

ChenShunHsu commented 11 months ago

Hello, I would like to ask about the .pt file of the goals task in task1. Are there any adjustments to the input size and any augmentation during pre-training?

tyb311 commented 11 months ago

The size of 1100 was cropped to 1024, then resized to 512 to train the model fastly. The size of 800 was cropped to 496. So 1100x800 to 512x496.

------------------ 原始邮件 ------------------ 发件人: @.>; 发送时间: 2023年12月19日(星期二) 凌晨3:21 收件人: @.>; 抄送: @.***>; 主题: [tyb311/TCCT] pretrained model setting (Issue #2)

Hello, I would like to ask about the .pt file of the goals task in task1. Are there any adjustments to the input size and any augmentation during pre-training?

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: @.***>

ChenShunHsu commented 11 months ago

How about training augment? Did I you used alb.CropNonEmptyMaskIfExists augment?