Xin Xin
Xin Xin
不包含
> when will we get the code it's almost an year i'm trying to run my research for that as a baseline i need to run this please share it...
I meet same issue in stage 1. The same issue in stage2 is solved by training enough steps for stage1 and stage2, however the issue in stage 1 is still...
然而RealisDance 也是只能看 不能推理和训练
@theFoxofSky 你提供的代码 连测试用例都没法制作呀
> @SzhangS 而且姿势绘制代码已经在pull request里面了,我只是还没有空merge 看见了 2 days ago 感谢开源
在预训练权重上微调的话,你微调的是model-state-110000.th吗?
嗯呢呗
> mm_grounding训练的方式里不会存在这种负样本情况 假如混合训练A和B数据集,A中有猫的框和标签,B中只有狗的框和标签,B中有猫但没有标记出来,mm_grounding_dino训练检测猫时,只会将B中的狗当作负样本? A中的无猫区域和B中的猫区域不会被当作负样本?