Skip to content

Latest commit

 

History

History
20 lines (18 loc) · 925 Bytes

README.md

File metadata and controls

20 lines (18 loc) · 925 Bytes

DataMining

Data mining project - image recognition

Idea

  • find clusters of similar images
  • find similar objects on the image and add tags of it

Technologies

CIFAR-10 image classification

  • Download the (Python version of the) CIFAR-10 dataset and extract it in a cifar-10 top-level directory (it is in the .gitignore, no worries).
  • Run python cifar_10_test.py:
    • reads the dataset
    • displays a random image and its label (from the first batch)
    • runs a classifier returning const class (5)
    • runs a classifier returning random classes
    • runs a classifier returning the most common class based only on a single byte (single color channel of a single pixel) - for all bytes (Best accuracy 0.18 for byte 2078)