Polyp Segmentation in Endoscopy Image

[Kaushal-11]

I propose to add a feature for Polyp Segmentation using an endoscopy image dataset. This feature aims to improve the detection and segmentation of polyps in endoscopic images, which is crucial for early diagnosis and treatment of colorectal cancer.

Kindly assign me the issue under the label of GSSoC'24.

Dataset : it contains various images by taking endoscoy.
Link : https://polyp.grand-challenge.org/CVCClinicDB/

Model : U - Net model is used for segmentation with using tensorflow library.

[github-actions]

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

[abhisheks008]

Hi @Kaushal-11 can you implement at least 3 models for this project?

[Kaushal-11]

Yes, i can implement different models also like DeepLab, FCN, ResNet.

[abhisheks008]

Cool, issue assigned to you @Kaushal-11

[Kaushal-11]

Hey, @abhisheks008
UNet model's training accuracy has reached 0.96, and the validation accuracy is at 0.93

[abhisheks008]

Hey, @abhisheks008 UNet model's training accuracy has reached 0.96, and the validation accuracy is at 0.93

What about the rest of the models?

[Kaushal-11]

I am working on it

[Kaushal-11]

Hey, @abhisheks008
2nd Model ResUnet model's training accuracy has reached 0.98, and the validation accuracy is at 0.95

And 3rd model is under working

[Kaushal-11]

Hey, @abhisheks008
I have completed my work on all models.

ResUnet model give accuracy 0.95,
DeepLabV3+ model give accuracy 0.94

[Kaushal-11]

I have done my task with 3 model,

so can you suggest me how to add this stuff in repository ?

[abhisheks008]

I have done my task with 3 model,

so can you suggest me how to add this stuff in repository ?

Clone/fork this repo, make the changes in the forked repo and push your code by a PR.

[Kaushal-11]

okay, I have done it.
Now check all the stuffs,
If you see any areas that need improvement please inform me.

[github-actions]

Hello @Kaushal-11! Your issue #626 has been closed. Thank you for your contribution!