Closed wangjuanjuan1216 closed 1 year ago
Absolutely yes, but you need to convert the annotations to corresponding formats, like txt format for YOLO series.
So after transferring the original labels, if I want to send a new model for training, do I also need to split the image into 800*800 patches like you, and fill the ignore area with black, so as to get the new label after preprocessing?
Yes
Hello author, can this sodad dataset you made be directly used in algorithms that are not part of the MMdetection integration? For example, directly into the yolov7 or yolov8 algorithm?