NYU-CV-Fall-2018

Assignment 2: Traffic sign competition

Requirements

1. Install PyTorch from http://pytorch.org

2. Run the following command to install additional dependencies. Pay attention that the version of torchvision is specifically 0.4.1

pip install -r requirements.txt

Training and validating your model

Run the script main.py to train your model.

• By default the images are loaded and resized to 244 x 244 pixels and normalized to zero-mean and standard deviation of 1. See data.py for the data_transforms.
• By default a validation set is split for you from the training set and put in [datadir]/val_images. See data.py on how this is done.

Evaluating your model on the test set

As the model trains, model checkpoints are saved to files such as model_x.pth to the current working directory. You can take one of the checkpoints and run:

python evaluate.py --model ensemble.pth --data [data_dir]

The code uses three checkpoint files. One for the DenseNet model, one for the MobileNet model and one for the Ensemble model. Download the .pth files from the below link and put them in the root folder before running main.py or evaluate.py.

Link to PTH files : https://drive.google.com/open?id=1mu37TaIeB4KFb1iSx2mBIfncl2ByDZFY