deep-residual-neural-network-design

Using PyTorch and TorchVision to develop and image classification pipeline via the design of a CNN architecture

(extreme gratitude and appreciation to Dr. Catia Silva for being an amazing professor)

Dependencies

1. Create a new conda environment ml_experts with the following dependencies.

conda create -n ml_experts python=3.7 torchvision cudatoolkit=11.3 pytorch=1.12.1=gpu_cuda* -c pytorch

conda install -c conda-forge matplotlib

conda install -c conda-forge imutils

conda install -c conda-forge tqdm

conda install -c anaconda scikit-learn

Installation

1. Clone the repo.

git clone https://github.com/UF-FundMachineLearning-Sp23/final-project-ml-experts.git

2. Setup and activate environment

conda activate ml_experts

Usage

Training

1. Put data_train.npy and labels_train.npy in the base directory. The shapes should be (270000, n_samples) and (n_samples, ) respetively.
2. python prepare_train_data.py
3. python train.py

How it works:

1. prepare_train_data.py creates images from data_train.npy and save them in logo_photos folder. Next it creates dataset/train/ and dataset/val/ folders and put the images according to VAL_SPLIT ratio.
2. train.py imports pretrained ResNet-50 model and finetune it on provided train and validation data. The model is saved at output/finetune_model.pth.

Note: If you want to change any parameters of training, please do it in utils/config.py. Here are some variables that you can change.

1. TRAIN_DATA : Name of your data file. (a .npy file)
2. TRAIN_LABELS : Name of your labels file. (a .npy file)
3. VAL_SPLIT : Train/Validation ratio. (default is 70/30)
4. IMAGE_SIZE : Input image size to model. (default is 224, must be multiple of 32)
5. FINETUNE_BATCH_SIZE : Training batch size. (default is 64)
6. EPOCH : Training epoch. (default is 100)
7. LR : Learning rate. (default is 0.001)
8. FINETUNE_MODEL : Directory and name for saving the model

Testing

1. Put data_test.npy and labels_test.npy in the base directory. The shapes should be (270000, n_samples) and (n_samples, ) respetively.
2. python test.py

How it works:

1. The test.py creates images from data_test.npy and saves them in dataset/test/ folder. Next it generates pred_list that contains predicted labels. It also prints a classification report.
2. If you trained and saved a model with different name then use: python test.py --path <path to your model>

Authors

1. Imran Nasrullah email
2. Nicholai Platonoff email
3. Adnan Abdullah email