Contribution.txt

Contributions:
---------------

All three of us have individually tried to work on most of the parts of the projects. All of us had inputs on data preparation, augmentation, building models, tuning parameters, evaluating models and presenting them. In order to achieve such an ambitious yet achievable analysis, we separated the work between all of us in the following way:

● Nhung Le (nhl256): 
--------------------
* Preparing train, validation and test files for 2-class models
* Implementing ResNet18, ResNet34, and customized CNN models for 2-class models
* Tuning hyper-parameters and optimization function for ResNet18, ResNet34, ResNet50, and customized CNN models
* Developing evaluation code and graphs for 2-class models

● B V Nithish Addepalli (bva212): 
-----------------------------
* Downloading the data and uploading it into HPC
* Preprocessing data for Inception v2 and ResNet50 models
* Implementing ResNet50, Inception v2 models for 2-class and 4-class models
* Implementing evaluation code and graphs for 4-class models

● Ravi Choudhary (rc3620): 
--------------------------
* Preparing train / test / valid csv files for 4-class models
* Pre-processing data for customized CNN and Resnet18 for 4-class models
* Investigating and transforming images for 4-class models
* Implementing customized CNN, Resnet18, bounding box models