Add Object detection models and replace existing models. (Restructuring Pt - 3)

mlpack / models

models built with mlpack

https://models.mlpack.org/docs

BSD 3-Clause "New" or "Revised" License

35 stars 41 forks source link

Add Object detection models and replace existing models. (Restructuring Pt - 3) #8

Open kartikdutt18 opened 4 years ago

kartikdutt18 commented 4 years ago

Hey everyone, This PR aims to remove existing examples and add model for object classification. This PR will help set structure for one of my next PRs regarding Darknet-19 and Darknet-53. Things that I are left to do :

Train the model.
Upload weights on mlpack.org and make changes in load to reflect the same.
Incorporate suggestions.
Add documentation for this model.

{There will be another PR for documentation of dataloader with a simple tutorial.txt}. Looking forward to your thoughts and suggestions. Thanks a lot.

kartikdutt18 commented 4 years ago

Adding a ToDo List here :

[x] Set up Training and Testing Script.
[ ] Host weights online.
[x] Add function to unzip files.
[x] By end of this PR, run all tests (instead of the single one).
[ ] Remove LSTM, Kaggle, and cnn etc.

kartikdutt18 commented 4 years ago

Trained all models, the tests pass locally. Let's add unzip function and host the weights online in the next two days. Fix the styling issues.

kartikdutt18 commented 4 years ago

Right, we need to link unzip function to everything, and Also upload mnist_test in mnist.tar in mlpack.org/datasets. Also host weights online. Once that is done, we can merge this after taking into account all suggestions. Also unzip function has been added without adding any other dependency. These messages are mostly to maintain a log of what needs to be done and can be ignored. Sorry about that. Regards.

birm commented 4 years ago

Should we mark this as a draft, or are you ready for review?

kartikdutt18 commented 4 years ago

I'll make one more commit today and then it will be ready for review.

kartikdutt18 commented 4 years ago

I am guessing that the invalid access in memory is maybe due to mlpack/mlpack#2314. Let see what this build outputs. Thanks a lot.

kartikdutt18 commented 4 years ago

Let's open 2 PR one for LeNet and one for utility function (because that is ready) and LeNet will have to wait for copy constructors I guess.

mlpack-bot[bot] commented 4 years ago

This issue has been automatically marked as stale because it has not had any recent activity. It will be closed in 7 days if no further activity occurs. Thank you for your contributions! :+1:

kartikdutt18 commented 4 years ago

Keep Open.