MobileNetv2 - Possible Network mismatch with Publication/PyTorch
Nobody has claimed this yet.
- Dominant language
- Jupyter Notebook
- Stars
- 9.8k
- Forks
- 1.6k
- PR merge metrics
- No merged PRs in 30d
Description
There seems to be a mismatch of the MobileNetv2 model (https://github.com/onnx/models/blob/master/vision/classification/mobilenet/model/mobilenetv2-7.onnx) and and the MobileNetv2 publication/Pytorch implementation:
https://arxiv.org/pdf/1801.04381.pdf
https://github.com/pytorch/vision/blob/master/torchvision/models/mobilenet.py
If you check Table 2 in the publication, the strides are set as follows [column s]:
2, 1, 2, 2, 2, 1, 2, 1, 1
I have the impression your onnx model has:
2, 1, 2, 2, 1, 2, 2, 1, 1
This increases the number of MAdds substantially from ~300M to over 430M.
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Compare the linked ONNX MobileNetv2 model with Table 2 of the publication and the linked torchvision implementation, focusing first on the reported stride sequence. Verify whether the discrepancy changes the MAdds from about 300M to over 430M; done means the model architecture and reported computation are reconciled or the mismatch is confirmed with evidence.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- pytorch
- Domain
- machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100