How akshat151/cs5330_deeplearning_based_covid19_detection works
Project Type and Purpose
• This is a deep learning project aimed at COVID-19 detection using imaging data.
• It leverages several convolutional neural network architectures (DenseNet, LeNet, Resnet, VGG) that appear to be implemented in individual Jupyter notebooks.
• The README explains that the project uses a conda environment (defined in “cs5330_project_env.yml”) and datasets from external sources, with notebooks for evaluation, preprocessing, and inference.
Analyzing the File Structure
• Top-Level Organization:
• Model Folders: “DenseNet”, “LeNet”, “Resnet”, and “VGG” each contain individual notebooks for model experimentation.
• Other key notebooks in the root include “Inference.ipynb” (plus associated PDF), “Preprocessing.ipynb”, and “evaluation.ipynb”.
• The environment’s configuration file (“cs5330_project_env.yml”) indicates dependency management.
Key Components
a. Data/External Resources:
• External relationships using dataset links (Kaggle and OneDrive) presented in the README.
• Data source as “Dataset Source” nodes.
b. Environment & Setup:
c. Preprocessing Module:
• This component is represented by “Preprocessing.ipynb,” where raw data is likely cleaned and formatted.
d. Model Components:
• Each model (DenseNet, LeNet, Resnet, VGG) gets its own block.
e. Inference & Evaluation:
• “Inference.ipynb” (with supporting documentation “Inference_Code.pdf”) and “evaluation.ipynb” as separate modules that use the pre-trained models to generate predictions and assess performance.
f. Data Flow and Interconnections:
Architectural Patterns & Design Principles
• The modular separation indicates a design similar to a pipeline (data ingestion/preprocessing → model selection/training/inference → evaluation).
• It is based on the “divide and conquer” idea where each model architecture can be developed, updated, or evaluated independently.