inceptionnext performance

Question

inceptionnext performance

suisui-su-crtl opened this issue 2 years ago · 16 comments

Hello, I tried to use inceptionnext as the encoder, but the experiment results are not ideal ( The performance is even worse than convnext, I think maybe I made a mistake), can you help me to see if there is a problem with my code?

Since the encoder is connected to the decoder at each layer, I need to extract the results of each layer.

original:

def forward(self, x):
    x = self.forward_features(x)
    x = self.forward_head(x)  # I removed this part of the code because of the encoder
    return x

I modified:

def forward(self, x):
    y = x
    y = self.stem(y)
    y = self.stages[0] (y)
    x1 = y
    
    y = self.stages[1] (y)
    x2 = y
    
    y = self.stages[2] (y)
    x3 = y
    
    y = self.stages[3] (y)
    x4 = y
    
    return x1, x2, x3, x4

Answer 1 · 2023-05-22T06:12:02.000Z

Hi, for the semantic segmentation task in the paper, we add SyncBN for the output of each stage, like
return syncbn1(x1), syncbn2(x2), syncbn3(x3), syncbn4(x4).

Answer 2 · 2023-05-22T09:05:42.000Z

Thank you very much for your help. My research area is semantic segmentation.

Do you mean to add SyncBN layer after the output of each layer?

I have some questions about SyncBN layer.

I don't see the use of SyncBN in code. Do I need to add it by myself?

I use a single card for training, search found that SyncBN can support multi-card training, syncbn is useful for my training? Or is it more appropriate to directly add a BN layer?

Looking forward to your reply, which is very important to me.

Answer 3 · 2023-05-23T13:04:39.000Z

Hi, yes, BN needs to add for the semantic segmentation task. For a single GPU, BN is equal to SyncBN. For multi-GPU training, please remember to use SyncBN.

Answer 4 · 2023-05-24T02:02:59.000Z

in instance segmentation task,I didn't use the BN for the output of each stage，and my result of experiment is bad.

Answer 5 · 2023-05-24T02:04:48.000Z

I don't know the reason.I use inceptionnext-tiny to replace resnet50, why the result is worse.

Answer 6 · 2023-05-24T11:46:41.000Z

Hi @116022017144 ,

Besides adding SyncBN for the output of each stage, remember to replace BN in InceptionNeXt with SyncBN.

Answer 7 · 2023-05-24T12:40:36.000Z

ok,thank you！Do all the BN is needed to be replaced?

Answer 8 · 2023-05-24T12:44:45.000Z

还有对每个stage输出做SyncBN，这样可以吗？self.stages = nn.Sequential(norm_layer(dims[0]))，还是只需要在声明形参那里这样 norm_layer=nn.SyncBatchNorm,就行

Answer 9 · 2023-05-24T12:48:34.000Z

这个类class MlpHead(nn.Module)，如果做分割任务的话，我看是没有用到的，其norm层不用管了吧？

Answer 10 · 2023-05-24T12:52:08.000Z

Hi @116022017144 , yes, I remove all MlpHead for semantic segmentation.

Answer 11 · 2023-05-24T13:00:55.000Z

I met the error. And my model support distrbuted training.

Answer 12 · 2023-05-25T14:03:26.000Z

Hi @116022017144 , yes, I remove all MlpHead for semantic segmentation.

Can you come up with a semantic segmentation version of inceptionnext when you have time? I would really appreciate it if you could!

I added BN layers after each output layer, but the accuracy is only slightly improved, similar to the performance of convnext. I'm having a hard time finding the problem.

Answer 13 · 2023-05-25T14:06:52.000Z

sorry,I use it in instance segmentation task, so I can't help you! Now I don't replace BN，because I met some errors

…

---Original--- From: ***@***.***> Date: Thu, May 25, 2023 22:03 PM To: ***@***.***>; Cc: ***@***.******@***.***>; Subject: Re: [sail-sg/inceptionnext] inceptionnext performance (Issue #10)

Answer 14 · 2023-05-25T14:20:58.000Z

I met the error. And my model support distrbuted training.

Hi, I use MMSeg for semantic segmentation. SyncBN code like this

from mmcv.cnn import build_norm_layer

norm_cfg=dict(type='SyncBN', requires_grad=True)

self.norm = build_norm_layer(norm_cfg, out_chs)[1]

Answer 15 · 2023-05-25T14:29:15.000Z

Hi @116022017144 , yes, I remove all MlpHead for semantic segmentation.

Can you come up with a semantic segmentation version of inceptionnext when you have time? I would really appreciate it if you could!

I added BN layers after each output layer, but the accuracy is only slightly improved, similar to the performance of convnext. I'm having a hard time finding the problem.

Hi, sorry that I am so busy recently and may have no time to clean the code. Following ConvNeXt, we report mIoU multi-scale testing in Table 5 in the paper, where InceptionNeXt significantly outperforms ConvNeXt. Besides, I also evaluate it with single scale, finding the performance of InceptionNeXt is similar to ConvNeXt. It seems InceptionNeXt is more in favor of multi-scale testing.

Answer 16 · 2023-05-25T14:58:46.000Z

Hi @116022017144 , yes, I remove all MlpHead for semantic segmentation.

Can you come up with a semantic segmentation version of inceptionnext when you have time? I would really appreciate it if you could!
I added BN layers after each output layer, but the accuracy is only slightly improved, similar to the performance of convnext. I'm having a hard time finding the problem.

Hi, sorry that I am so busy recently and may have no time to clean the code. Following ConvNeXt, we report mIoU multi-scale testing in Table 5 in the paper, where InceptionNeXt significantly outperforms ConvNeXt. Besides, I also evaluate it with single scale, finding the performance of InceptionNeXt is similar to ConvNeXt. It seems InceptionNeXt is more in favor of multi-scale testing.

Ok, I can understand. Thank you very much for your help many times. I will read the paper in detail again to find the answer.