search

abhisheks008 / DL-Simplified

Deep Learning Simplified is an Open-source repository, containing beginner to advance level deep learning projects for the contributors, who are willing to start their journey in Deep Learning. Devfolio URL, https://devfolio.co/projects/deep-learning-simplified-f013
https://quine.sh/repo/abhisheks008-DL-Simplified-499023976
MIT License
360 stars 300 forks source link

[Model Enhancement]: Chocolate Classification using DL #553

Closed Arihant-Bhandari closed 4 months ago

Arihant-Bhandari commented 4 months ago

Deep Learning Simplified Repository (Proposing new issue)

:red_circle: Project Title : [Model Enhancement]: Chocolate Classification using DL
:red_circle: Aim : enhancing custom CNN model
:red_circle: Dataset : https://www.kaggle.com/datasets/siddharthmandgi/chocolate-classification
:red_circle: Approach : adding attention as overhead to CNN architecture and using ELU and NADAM in activation and optimizer.

📍 Follow the Guidelines to Contribute in the Project :

:red_circle::yellow_circle: Points to Note :

:white_check_mark: To be Mentioned while taking the issue :

All the best. Enjoy your open source journey ahead. 😎

Arihant-Bhandari commented 4 months ago

@abhisheks008 hi, can i have a go at this, i feel attention can help boost accuracy for CNN. If i find my work substantial and satisfactory i will PR, if not i will close the issue on my end.

thank you for your time.

abhisheks008 commented 4 months ago

Are you planning to enhance this project, https://github.com/abhisheks008/DL-Simplified/tree/main/Chocolate%20Classification%20using%20DL right?

Arihant-Bhandari commented 4 months ago

yes @abhisheks008 , the CNN is at 0.86 currently, i think it can be boosted to 0.94 ish with attention, if i do end up achieving this i will start PR.

abhisheks008 commented 4 months ago

Follow the same representation like you did in the previous issue. Go ahead with this. Assigned to you @Arihant-Bhandari

Arihant-Bhandari commented 4 months ago

@abhisheks008 hi, i think i have completed the project, it has been boosted as i expected, but i want to send over the file without the actual PR for you to see before we make things official with PR.

is it possible, like with an email or on discord ?

abhisheks008 commented 4 months ago

@abhisheks008 hi, i think i have completed the project, it has been boosted as i expected, but i want to send over the file without the actual PR for you to see before we make things official with PR.

is it possible, like with an email or on discord ?

Chill dude, make a PR, if any changes are required you can update them flawlessly.

Arihant-Bhandari commented 4 months ago

i had 2 possible solutions actually both are same implementation its just model structuring differences. how about i sent them over , you can have a look at them @abhisheks008.

abhisheks008 commented 4 months ago

You can sent them over abhishek.opensource@gmail.com. My suggestion is put both of them together in the .ipynb file, why not showcase a comparison. That'll be insightful.

Arihant-Bhandari commented 4 months ago

cool we can do this, i will push them and send a PR in, thanks for listening @abhisheks008

Arihant-Bhandari commented 4 months ago

@abhisheks008 i have sent the files over email, pls check, also i noticed someone was working on the README file, issue #526, should i wait for them before updating readme with metrics like i did in the previous project or do it on my own ? i can hand them metrics and image link if they are working on it.

abhisheks008 commented 4 months ago

@abhisheks008 i have sent the files over email, pls check, also i noticed someone was working on the README file, issue #526, should i wait for them before updating readme with metrics like i did in the previous project or do it on my own ? i can hand them metrics and image link if they are working on it.

@abhisheks008 i have sent the files over email, pls check, also i noticed someone was working on the README file, issue #526, should i wait for them before updating readme with metrics like i did in the previous project or do it on my own ? i can hand them metrics and image link if they are working on it.

@abhisheks008 i have sent the files over email, pls check, also i noticed someone was working on the README file, issue #526, should i wait for them before updating readme with metrics like i did in the previous project or do it on my own ? i can hand them metrics and image link if they are working on it.

Let's wait for that issue to solved.

Arihant-Bhandari commented 4 months ago

hi @abhisheks008 , have you reviewed the codebase, which should i be pushing as PR ?

abhisheks008 commented 4 months ago

I checked the code and found that you have overfitted the train dataset as a result it is giving an accuracy of 100% which is an eye catching event. Try to increase the test dataset portion.

Arihant-Bhandari commented 4 months ago

hi @abhisheks008 im starting my PR i think i am on the right track here, pls review the codebase, if any changes needed i will be more than glad to do things over. thank you for your patience.