[Model and README Enhancement] Mushroom Classification using Deep Learning

[Arihant-Bhandari]

Deep Learning Simplified Repository (Proposing new issue)

🔴 Project Title : Mushroom Classification using Deep Learning

🔴 Aim : Bettering models by adding keras implemented CNN and CNN with attention models

🔴 Dataset : https://www.kaggle.com/datasets/lizhecheng/mushroom-classification

🔴 Approach : The original author used MAE as his loss value and implemented models, i will be trying out keras based CNN models with categorical cross entropy loss turning the problem into multiclass classification.

---

📍 Follow the Guidelines to Contribute in the Project :

• You need to create a separate folder named as the Project Title.
• Inside that folder, there will be four main components.
  • Images - To store the required images.
  • Dataset - To store the dataset or, information/source about the dataset.
  • Model - To store the machine learning model you've created using the dataset.
  • requirements.txt - This file will contain the required packages/libraries to run the project in other machines.
• Inside the Model folder, the README.md file must be filled up properly, with proper visualizations and conclusions.

---

🔴🟡 Points to Note :

• The issues will be assigned on a first come first serve basis, 1 Issue == 1 PR.
• "Issue Title" and "PR Title should be the same. Include issue number along with it.
• Follow Contributing Guidelines & Code of Conduct before start Contributing.

---

✅ To be Mentioned while taking the issue :

• Full name : Arihant Bhandari
• GitHub Profile Link : https://github.com/Arihant-Bhandari
• Email ID : [email protected]
• Participant ID (if applicable): Arihant_Bhandari (eternal_insight)
• Approach for this Project : adding 2 model implementations under multiclass classification
• What is your participant role? (Mention the Open Source program)
  GSSoC Contributor 2024

---

Happy Contributing 🚀

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

[github-actions]

Thank you for creating this issue! We'll look into it as soon as possible. Your contributions are highly appreciated! 😊

[Arihant-Bhandari]

hi @abhisheks008 please assign me this, i will also be working on the project idea i asked you about, as soon as i finish with some basic results i will create issue for it as well.

[abhisheks008]

You have already did this issue, #643

[Arihant-Bhandari]

hi @abhisheks008 just realized i set the wrong title for this, i was hoping to work on Mushroom Classification problem.

[abhisheks008]

The existing project is consisting CNN and Inception for this project, what are models you are planning for this to enhance the accuracy? Please be specific with the architecture names/models.

[Arihant-Bhandari]

i was hoping to add in a CNN-Attention model based on keras for this, also i think the original author for this used MAE as loss for some models, i was thinking about turning this into a multiclass classification model using softmax and categorical_crossentropy.

[abhisheks008]

Implement at least 2 more models to get a level 2 tag, otherwise it will be considered as level 1.

Assigning this issue to you @Arihant-Bhandari

[Arihant-Bhandari]

hi @abhisheks008 i wanted to know how i could turn this issue into level3, from what i gather based on the previous work, the dataset yields very low results , and i myself face similar issues, my best went to about 0.3 % accuracy on baseline CNN, so what i wanted to ask is, if i were to colour train the model, say 3 models whose outputs are voted on for each colour channel as part of my submission alongside 2-4 pretrained models, would this qualify for a level3 contribution ?

[abhisheks008]

hi @abhisheks008 i wanted to know how i could turn this issue into level3, from what i gather based on the previous work, the dataset yields very low results , and i myself face similar issues, my best went to about 0.3 % accuracy on baseline CNN, so what i wanted to ask is, if i were to colour train the model, say 3 models whose outputs are voted on for each colour channel as part of my submission alongside 2-4 pretrained models, would this qualify for a level3 contribution ?

Push your code and let me review it. Will definitely let you know if that qualifies for the level 3.

[Arihant-Bhandari]

hi @abhisheks008 i can send in preliminary work, i devised a custom data collection mechanism since the original data owner noted that there are issues with the daatset and people working on EDA concurred with following issues: the images had some duplicates, some dark and some grayscaled images when the dataset was supposed to be purely RBG.

alongside this i am also sending in 5 model's work based on how the original author of the repo did his work: i have implemented RESNET50, VGG16, Xception, DenseNet and Inception-ResNet-v2.

in addtion i did a baseline CNN model in keras , which had highest accuracy in all cases as part of followup towards attention based model.

[abhisheks008]

hi @abhisheks008 i can send in preliminary work, i devised a custom data collection mechanism since the original data owner noted that there are issues with the daatset and people working on EDA concurred with following issues: the images had some duplicates, some dark and some grayscaled images when the dataset was supposed to be purely RBG.

alongside this i am also sending in 5 model's work based on how the original author of the repo did his work: i have implemented RESNET50, VGG16, Xception, DenseNet and Inception-ResNet-v2.

in addtion i did a baseline CNN model in keras , which had highest accuracy in all cases as part of followup towards attention based model.

Push all your codes together.

[github-actions]

Hello @Arihant-Bhandari! Your issue #651 has been closed. Thank you for your contribution!