-
Notifications
You must be signed in to change notification settings - Fork 4.2k
-
Notifications
You must be signed in to change notification settings - Fork 4.2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
If my custom dataset are all 300X300, or 600X600, is there necessary to train a model use image-size1280? #2045
Comments
@JacksonWang YOLO model resizes the images to it's network dimensions before starting the training. So no matter what image sizes you have, it will be resized to it's network size( 640 or 1280). Since you have images of size less than 640x640, try any of the following smaller models with network size of 640 for the training- |
@dsbyprateekg Thanks for your suggestion. Is it not recommended to use 1280 for training because if the size of the original photo is enlarged, it may cause distortion or other reasons, making the training results even more abnormal? |
@JacksonWang selection of models always depend on the dataset complexity. If you have very small objects (less than 16x16 pixels for example) to detect in your 600x600 images, you should train with bigger model like YOLOv7-w6. |
@dsbyprateekg Thank you for such a detailed explanation. |
Hello, I have a question. I am currently starting to learn the model training of yoloV7, but there is a problem that has troubled me for a long time. If I have my own data set to train, but its size (WxH) is all If it is only 300x300 or 600x600, is it necessary to use image-size 1280 for training? What is the reason for this?
The text was updated successfully, but these errors were encountered: