Shanghua Gao
Shanghua Gao
> Hi @gasvn , thanks for your contributions. Please add docstring and type hints for your added files. I have added docstring and type hints.
Is there any further update about this pull request?
Is there anything I can do to make some progress? Thanks~
https://github.com/Res2Net/Res2Net-PretrainedModels/blob/master/res2net_v1b.py 你可以直接参照我们的代码。我们的代码也是根据resnet50改的。从你给的信息我无法判断出来，一般原因可能是参数没传对。
> width = int(math.floor(planes * (baseWidth/64.0)))您好，这一行代码我一直没太弄明白，请问有人知道么 这只是为了方便跟resnet对齐参数量写的一行代码。
Res2Net has no aspp module. If you do the segmentation task with Res2Net, only the weights in Res2Net are initialized with ImageNet pretraining, the weights in ASPP can be initialized...
res2net_v1b is a improved version of res2net, which follows the structure of resnet_v1d and has better training scheme. Res2NeXt is based on the ResNeXt. res2net_v1b has the best performance.
> res2net_v1b在res2net基础具体上修改了哪些呢 我们follow resnet_v1b修改了input的conv和downsample block
We adjuct the width of models with each scale to make the overall parameters and FLOPS constant with resnet-50. It's easy to keep the Flops as the overall structure is...
我们暂时没有提供，抱歉。