is it possible to train a detection model that can detect the skewed/tilt word/license plate?(重新訓練det模型以檢測出歪斜字樣或車牌號碼是可行的嗎?) #14243
t109598032
started this conversation in
General
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
以下圖車牌號碼為例，以en-PPOCRv3-det檢測出來的結果如左上角，這個綠框會趨近直角四邊形，但因為是俯視拍攝，該框內容經仿射變換會變成有點扭曲、歪斜，在某些案例可能導致辨識度不佳，所以想問問是否能重新訓練一個det模型，把標註框改成貼合字體(如左下角)，使得檢測出來的框就是趨近這種形狀，之後接仿射變換能較為正楷，使得辨識更準確?這個想法在目前det網路架構上重新訓練是可行的嗎?
Beta Was this translation helpful? Give feedback.
All reactions