DliHjiang

How are region-level descriptions obtained?

1

Thanks for your great work！ In your paper, the label of the semantic classification branch is mask cropped embedding obtained by CLIP, then how is the GT of the caption...

请问您有复现过Distilling Object Detectors with Fine-grained Feature Imitation论文中的方法吗？

2

我最近在带FPN的faster rcnn中应用论文中的方法，但是效果很差，比学生的基线都低，您有在带fpn的faster rcnn上应用过该方法吗？如果是在带fpn的faster rcnn上，gt要如何分配到不同的level上呢？

when will the training data be released？

2

Thanks for your great work！ Now I want to conduct some experiment of RAM，but the training data have not been released. May I ask when will the data set be...

can‘t ’get 51.8 AP on coco use the source code

1

Thanks for your great work！ I have used the source code to reproduce the performance of mask_rcnn_vitbase，I used the original "mimdet_vit_base_mask_rcnn_fpn_sr_0p5_800_1333_4xdec_coco_3x.py" config with 64 4090D GPU, But I have just...