Join GitHub today
GitHub is home to over 50 million developers working together to host and review code, manage projects, and build software together.
Sign upVGG Classification network #2083
Conversation
|
What is https://en.wikipedia.org/wiki/VGG and why does the reader need to guess? |
|
It's the implementation of https://arxiv.org/abs/1409.1556 a classification network |
TravisBuddy
commented
Jun 8, 2020
|
Hey @jeffin07, TravisBuddy Request Identifier: 7110f230-a9a7-11ea-9161-4f77e253115e |
TravisBuddy
commented
Jun 8, 2020
|
Hey @jeffin07, TravisBuddy Request Identifier: 28f2ebb0-a9a8-11ea-9161-4f77e253115e |
The problem is that the URL does not contain the acronym VGG and the code comments do not explain what VGG stands for. It is going to confuse the reader to have class without knowing what the class name stands for. |
TravisBuddy
commented
Jun 8, 2020
|
Hey @jeffin07, TravisBuddy Request Identifier: 4b310660-a9aa-11ea-9161-4f77e253115e |
| return f"VGG for feature extraction : \n {self.net}" | ||
|
|
||
|
|
||
| cfg = { |
This comment has been minimized.
This comment has been minimized.
deadshotsb
Jul 3, 2020
Member
This repository is based on learning purpose. So, it will be better if you could provide an explanation of the usage of cfg parameter.
|
@jeffin07 You can change the VGG to VGG-16/19 term as they are mainly the standard ones and people refer to them mostly. |
|
@cclauss VGG is a CNN classification, you can get it by using VGG-16/19. There is one problem though, the PR contains the model defination using pytorch and no insights to the layers(their functionality) but am not so sure that if you could call this as a work in algorithms. |
|
|
VGG(Visual Geometry Group) or also OxfordNet. Similarly there are AlexNet or ResNET, etc |
|
Thanks, please change the filename to |
|
@cclauss Actually VGG is often used and using a VGG-16/19 model and renaming the file as vgg_16 / vgg_19 would be beneficial, like using kmp instead of Knuth Morris Pratt. |
|
In this repo we call it Knuth Morris Pratt instead of kmp. We do that very deliberately because we are trying to teach the readers of our code. We avoid unexplained acronyms because we want our readers to understand the origins and nomenclature of these algorithms. The 16 and 19 are config choices that have proven to be effective but they are not the algorithm. https://www.robots.ox.ac.uk/~vgg/research/very_deep |
jeffin07 commentedJun 8, 2020
Describe your change:
Added a deep-learning method for image classification #1809 using pytorch
Checklist:
Fixes: #{$ISSUE_NO}.