How sdthakur/kidneydiseaseclassification works
KidneyAI is a research prototype for classifying CT images as normal or tumor. Its user workflow accepts an image through a Flask web/API entry point, decodes it, runs multi-crop model inference, and returns a class and confidence values. A separate training workflow orchestrates ingestion, model preparation, training, and evaluation. The README documents additional console features and MLflow tracking; those subsystems are represented, though their implementation details were not sampled. Edges are limited to relationships shown in excerpts or documented in the README.
Main entry points: app.py Flask routes · main.py Training orchestrator · configuration.py Configuration manager · config_entity.py Stage config types
Components 19
Inference and console
- Flask routes
app.py - Image decoder
src/cnnClassifier/utils/common.py - Multi-crop inference
src/cnnClassifier/pipeline/prediction.py - Trained weights
model/model.h5 - Diagnostic console
templates/index.html
Training workflow
- Training orchestrator
main.py - Ingestion stage
src/cnnClassifier/pipeline/stage_01_data_ingestion.py - Preparation stage
src/cnnClassifier/pipeline/stage_02_prepare_base_model.py - Training stage
src/cnnClassifier/pipeline/stage_03_model_training.py - Evaluation stage
src/cnnClassifier/pipeline/stage_04_model_evaluation.py
Model lifecycle
- Dataset ingestion
src/cnnClassifier/components/data_ingestion.py - VGG16 preparation
src/cnnClassifier/components/prepare_base_model.py - Model training
src/cnnClassifier/components/model_training.py - Metrics evaluation
src/cnnClassifier/components/model_evaluation_mlflow.py - MLflow tracking
Configuration
- Configuration manager
src/cnnClassifier/config/configuration.py - Stage config types
src/cnnClassifier/entity/config_entity.py - Shared utilities
src/cnnClassifier/utils/common.py
Other components
- Browser / API client
Connections 28
- Browser / API client → Flask routes: sends requests
README.md - Flask routes → Diagnostic console: renders page
app.py - Flask routes → Image decoder: decodes image
app.py - Flask routes → Multi-crop inference: invokes inference
app.py - Multi-crop inference → Trained weights: loads weights
src/cnnClassifier/pipeline/prediction.py - Flask routes → Browser / API client: returns prediction
README.md - Browser / API client → Training orchestrator: requests training
- Flask routes → Training orchestrator: launches runner
app.py - Training orchestrator → Ingestion stage: runs stage
main.py - Training orchestrator → Preparation stage: runs stage
main.py - Training orchestrator → Training stage: runs stage
main.py - Training orchestrator → Evaluation stage: runs stage
main.py - Ingestion stage → Configuration manager: gets config
src/cnnClassifier/pipeline/stage_01_data_ingestion.py - Ingestion stage → Dataset ingestion: runs ingestion
src/cnnClassifier/pipeline/stage_01_data_ingestion.py - Preparation stage → Configuration manager: gets config
src/cnnClassifier/pipeline/stage_02_prepare_base_model.py - Preparation stage → VGG16 preparation: prepares model
src/cnnClassifier/pipeline/stage_02_prepare_base_model.py - Training stage → Configuration manager: gets config
src/cnnClassifier/pipeline/stage_03_model_training.py - Training stage → Model training: runs training
src/cnnClassifier/pipeline/stage_03_model_training.py - Evaluation stage → Configuration manager: gets config
src/cnnClassifier/pipeline/stage_04_model_evaluation.py - Evaluation stage → Metrics evaluation: runs evaluation
src/cnnClassifier/pipeline/stage_04_model_evaluation.py - Configuration manager → Shared utilities: reads YAML
src/cnnClassifier/config/configuration.py - Configuration manager → Stage config types: builds configs
src/cnnClassifier/config/configuration.py - Dataset ingestion → Stage config types: uses config
src/cnnClassifier/components/data_ingestion.py - VGG16 preparation → Stage config types: uses config
src/cnnClassifier/components/prepare_base_model.py - Model training → Stage config types: uses config
src/cnnClassifier/components/model_training.py - Metrics evaluation → Stage config types: uses config
src/cnnClassifier/components/model_evaluation_mlflow.py - Metrics evaluation → Shared utilities: saves metrics
src/cnnClassifier/components/model_evaluation_mlflow.py - Metrics evaluation → MLflow tracking: logs metrics
README.md