issue

[PlutoNov]

When I plugged convnext into Faster-Rcnn, the effect was not ideal. My dataset is a bit small, with just over 2,000 training images and more than 700 test images, so I wanted to ask if the training accuracy is low because the dataset is too small.

[JACYI]

We are very sorry, we only tested the baseline performance of the ConvNeXt backbone in remote sensing image classification task and did not conduct research on the object detection dataset.