sunAttribute

Multi-label Classification for Sun Attribute Dataset

How to run:

1. Download the dataset from https://cs.brown.edu/~gen/Attributes/SUNAttributeDB_Images.tar.gz
2. Make a dir named "images" in "data/" and put the images in it
3. Run train.py to train a model (default is vgg16-based model)
4. Run test.py to test your trained model and get recall and precision

Code Structure:

• train.py: a script for training
• test.py: a script for testing
• classifier/dataset: parse the SUN Attribute Database (In most cases you do NOT need to make changes to it)
• classifier/models: different base models, including reset and vgg
• classifier/utils: some basic codes, including metrics, data transformer, data pre-processer and so on (In most cases you do NOT need to make changes to it)
• classifier/trainer.py: a trainer for training (In most cases you do NOT need to make changes to it)
• calssidier/evaluator.py: an evaluator for evaluation and testing (In most cases you do NOT need to make changes to it)

Some Instructions:

• `You can change parameters and settings by making modifications to train.py & test.py
• `Try different models by chaging classifier/models.py
• `Please provide me with results higher than my baseline model