dreamwish1998

@jeya-maria-jose If I plan to use 256*256 size images, do I need to modify the modules in the network? Which parts should be changed? Thanks!

> The support images do not include the query image. I use the 1-way 10-shot training as an example. In the base training stage, there are 10 support images and...

If the number of target classes for the task of CT segmentation is not 14, such as 1 or 2, can I still use your pretrained model? If yes, what...

As above, can you elaborate on the relevant settings?

> 谢谢你。[这些设置与其他项目（ https://github.com/ZhangGongjie/SAM-DETR](https://github.com/ZhangGongjie/SAM-DETR) ）基本相同。 Thanks for your reply, I would also like to ask what is the most suitable batch size for this model. How much GB of GPU memory...

I meet the same problem! Does anyone have the solution? @scsunjing, did you solve it? Is this build/box step necessary in the project?

> @junxiaopang1998 I change all "long" to "int64_t" in nms.cpp to install it successfully on win10. if you install it succeessfully, it will faster when train.py. if you test succeessfully,...

> @junxiaopang1998 my train test.py ran successfully, I always get out of memory error when set net.use_mask = True at test.py,can you give me the results of the test.py @scsunjing...

> no matter how many GPUs(11GB each) I use, I always get out of memory error when set net.use_mask = True at testing time, how can I solve this? @LXYTSOS...

> 我们按照之前的工作（TFA、ICML 2020）准备数据集（https://github.com/ucbdrive/few-shot-object-detection/tree/master/datasets）。请参阅下面的 coco 文件夹概述。 > > ``` > coco/ > annotations/ > trainvalno5k.json > instances_val2014.json # same as the 5k.json > ... # few-shot training examples > trainval2014/ >...