Skip to content

Repository for training and evaluating various CNN based Classification Models

License

Notifications You must be signed in to change notification settings

aman-chauhan/places365-nets

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Places365-Nets

Requirements

UNIX Packages

Packages Purpose
wget, tar Download and Extract Places365 Easy Format Dataset

Python Packages

Package Version
python 3.6
numpy 1.15.4
progressbar 2.5
imageio 2.4.1

Instructions to Train and Evaluate the Model

  1. ./download_places365 - Command to Download and Extract the Places365 Dataset.
  2. Clean the dataset and remove grayscale images.
    • python clean_places365.py train - Clean the training dataset
    • python clean_places365.py val - Clean the validation dataset

Files

  • download_places365 - Executable script to download and extract the Places365 dataset.
  • clean_places365.py - Script to clean the dataset.
  • classes.txt - Sorted list of classes in Places365.

Folders

  • places365_standard - Created by download_places365
    • train.txt - List of path to files in training set
    • val.txt - List of path to files in validation set
    • train - Contains 365 scene category partitions, with 900 images per category
      • airfield
      • airplane_cabin
      • airport_terminal
      • ... 362 more folders
    • test - Contains 365 scene category partitions, with 100 images per category
      • airfield
      • airplane_cabin
      • airport_terminal
      • ... 362 more folders