namndh/texture-classification

Texture Classification using Deep Learning

Technique 1:

1. Related research, papers:
   • Model with SAE and using Texture Descriptors as vectorizers
2. Issues:
   • Paper proposes 27 features based GLCM but can not find detailed documents
   • Can not implement sucessfully local spectral histogram extractor
3. to-do:
   • Solve the issues
   • Make comparision with technique 2

Technique 2:

1. Related research, papers:
   • ResNets
2. Issues:
   • Dataset with no test set. Hard to divide dataset to get a random testset.
3. to-do:
   • Solve the issues
   • Make comparision with technique 1