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
For this issue, I have to include the image in it.
I used DBNet for image detection and found a problem. Whether it's a larger backbone such as Resnet18 or a lightweight backbone called Mobilenetv3, there may be situations where the characters on the right side of the image cannot be detected,As shown in the following figure.
As shown above, the characters on the right are not segmented using a heat map, but the characters on the left can be segmented, which makes me very confused. Thinking that your experience is definitely better than mine, I would like to ask what are the reasons for this situation. By the way, initially to address this issue, when training for image preprocessing, when the image width was less than the width of the training image, 0 pixel values were filled in on the right side of the image, with a 50% probability of each filling 0 pixel values on the left and right sides, but the effect still showed right side detection.Do you have any suggestions?
This discussion was converted from issue #11208 on June 10, 2024 03:11.
Heading
Bold
Italic
Quote
Code
Link
Numbered list
Unordered list
Task list
Attach files
Mention
Reference
Menu
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
-
For this issue, I have to include the image in it.


I used DBNet for image detection and found a problem. Whether it's a larger backbone such as Resnet18 or a lightweight backbone called Mobilenetv3, there may be situations where the characters on the right side of the image cannot be detected,As shown in the following figure.
As shown above, the characters on the right are not segmented using a heat map, but the characters on the left can be segmented, which makes me very confused. Thinking that your experience is definitely better than mine, I would like to ask what are the reasons for this situation. By the way, initially to address this issue, when training for image preprocessing, when the image width was less than the width of the training image, 0 pixel values were filled in on the right side of the image, with a 50% probability of each filling 0 pixel values on the left and right sides, but the effect still showed right side detection.Do you have any suggestions?
Beta Was this translation helpful? Give feedback.
All reactions