Skin Cancer Image Classification

Skin Cancer Image Classification - Data Every Day

Рет қаралды 3,213

3 жыл бұрын

Hi guys, welcome back to Data Every Day!
On today's episode, we are looking at a dataset of skin lesion images and trying to predict the type of lesion for a given image. We will be using a TensorFlow convolutional neural network to make our predictions.
Here is a link to the Kaggle dataset:
www.kaggle.com/kmader/skin-ca...
And here is a link to my notebook from the video:
www.kaggle.com/gcdatkin/skin-...
Thanks so much for watching! If you enjoyed today's episode, be sure to subscribe and hit the bell for more content!
See you all tomorrow! :)

Пікірлер: 11

@AdityaThakurXD 3 жыл бұрын

Great work Sir. Thank you for this. 🌟

@gcdatkin 3 жыл бұрын

No problem! :)

@thanhquang9514 3 жыл бұрын

Do you think it is appropriate to use bayes for this dataset

@devloper_hs 3 жыл бұрын

How to add data augmentation , is data augmention on minor class only ok , if yes how to use with Image data generator And can I simply use ImageDataGenerator for data augmentation

@gcdatkin 3 жыл бұрын

For image recognition, it's not a bad idea to augment the whole dataset, as it provides a good challenge for the model. ImageDataGenerator can be used as a simple way to augment the images. Just specify which transformations you want when initializing the ImageDataGenerator object and they will be randomly applied to images that are fed through the generator. You can find a complete list of available transformations here: www.tensorflow.org/api_docs/python/tf/keras/preprocessing/image/ImageDataGenerator

@viane123456 3 жыл бұрын

After getting all codes how can I find whether It has skin cancer or not and which type of cancer?

@ishmamurrahman9425 3 жыл бұрын

The label mapping should be 0:akiec and 4:nv

@gcdatkin 3 жыл бұрын

Ah, thank you for clarifying. Where did you find that information?

@ishmamurrahman9425 3 жыл бұрын

@@gcdatkin I'm doing my undergrad thesis on HAM10000, every research paper maps it so. Also the Kaggle data set description lists the lesions in this order.

@gcdatkin 3 жыл бұрын

Thanks!

@lotfiamr8433 Жыл бұрын

@@ishmamurrahman9425 hello sir did you work on this project please ? I think that his model is overfiting