Closed xunkaixin closed 7 years ago
Is this for training, or for detection ?
training and detection
The images are resized before feeding to the network. But during the process, features might become small. Other than this there should not be an issue.
thank you
In this case would the output (box coordinates) of the network during detection be on the resized image of scaled to the original image?
In the current system, all coordinates are relative, not absolute and hence resizing is not an issue. Please see arapaho tests for an example of how this works.
Thank you again! I will have a look into it tomorrow.
On 25 Apr 2017 4:42 pm, "Prabindh Sundareson" notifications@github.com wrote:
In the current system, all coordinates are relative, not absolute and hence resizing is not an issue. Please see arapaho tests for an example of how this works.
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/prabindh/darknet/issues/25#issuecomment-297072713, or mute the thread https://github.com/notifications/unsubscribe-auth/AKRZ-HbGytgbpvfGDVolKZmmyNzifMPYks5rzhRygaJpZM4NC7rL .
the width and height are 416 in .cfg file, but my labeled image size is 1280*720, did i need cut the labeled image? or some other suggestion? thank you