Skip to content

Latest commit

 

History

History
19 lines (13 loc) · 944 Bytes

README.md

File metadata and controls

19 lines (13 loc) · 944 Bytes
Raw

ImageClassification

Running instructions for training the models

  1. Extract dl2021-image-corpus-proj into resource/data/original (this is already done if you download the google drive zip)
  2. Extract image-test-corpus-139ArJI into resource/data/original (this is already done if you download the google drive zip)
  3. Run python src/data_wrangling/generate_dataframe.py to get the train.csv
  4. Run python src/data_wrangling/training.py to train the CNN2 model

Location of adjustable parameters

  • src/utils/image_loader.py -- adjust (a) transforms
  • src/training.py -- adjust (b) model and (c) optimizer

Other scripts

  • src/error_analysis.py is used to see the images and their actual and predicted labels using our models
  • src/testing.py is used to predict the images in the testing folder and produce a test.csv file
  • src/plotting/label_plots.py creates some data exploration plots which we used in the report