search

akashxg / Mammogram-Image-Classifier

breast cancer histopathological image classification using TensorFlow
15 stars 0 forks source link

What data is used in your sourcecode? Could you provide the data you used? Thanks #1

Closed guyucowboy closed 6 years ago

guyucowboy commented 6 years ago

Hi, akashxg: Great work! Thanks for sharing your codes. I have a question to ask you. What data is used in your sourcecode? I found the the code extensions = ['jpg', 'jpeg', 'JPG', 'JPEG'] in the function create_image_lists in retain.py which indicated the data format is jpg. But I download the DDSM data as you said, the data format of the DDSM is LJPEG. Could you provide the data you used? Thanks a lot! Best regards Gu Yu

akashxg commented 6 years ago

Hey Gu Yu:

I used software to convert the LJPEG images into a JPG format. Here are the images that I used: https://www.dropbox.com/s/zj0pg2pymir7dji/mammogram_photos.zip?dl=0

Best, Akash

guyucowboy commented 6 years ago

Hi, akashxg: I have downloaded your images. Thank you very much! I have another question to ask you. What does "bottleneck values for training images" mean? Is there the documention for the project in detail? Thanks again. Best regards Gu Yu

akashxg commented 6 years ago

Bottlenecks are the layers in a network, in this case the ImageNet model, that compress feature representations to best fit in the available space.

Check out this tutorial & this doc: https://codelabs.developers.google.com/codelabs/tensorflow-for-poets/#3

http://www.cs.unc.edu/~wliu/papers/GoogLeNet.pdf

Best, Akash