Skip to content

Latest commit

 

History

History
48 lines (28 loc) · 1.84 KB

readme.md

File metadata and controls

48 lines (28 loc) · 1.84 KB

Read Me - Shared information

Share every detail and information that all we need to know here.

!!! it would be good to move the main demo ('demo.m') here, the rest keep in 'code' folder.

Folder Description :

  • Code : ?
  • Data : ?
  • features/caffe : Extracted deep-features from caffe. 'listing.m' keeps the index of images name. 'img_feats_final.mat' file contains feats matrix. each feat-vector represents extracted deep-feature for an individual image.
  • features/gist : Extracted gist scene features. 'gist.mat' stores the index of images name in 'listing' variable and 'gists' feature vectors. each gist-vector represents extracted gist feature for an individual image.

Code Description :

  • startup.m : initialization (i.e : include folders, tools, libraries)


NOTE: In the following modules, maybe you need first modify images dataset path or annotation table files

1- TO annotate Images (to normal/abnormal/noisy/...):

  • First run 'startup.m'.

  • Then 'annotation'.

2- TO annotate Images Normal/Abnormal Regions (to annotate attributes):

  • First run 'startup.m'.

  • Then 'attribute_annotation'.


Imput: Input attribute_annotation would be the output table of annotate image in step No.1 (cell array including image_name and tags
Output: Output is an XML file that stores image_names, tags, and annotatited points

3- TO extract Abnormality patches (Manual):

  • First run 'startup.m'.

  • Then 'attribute_annotation_abnormalbox'.


Imput: Input attribute_annotation_abnormalbox would be the output XML file of attribute_annotation in step No.2
Output: Output is an cell table that stores image_names, and cordinate of patches around the center of abnormality points