jagratadeb/pneumonia-detector

How jagratadeb/pneumonia-detector works

Pneumonia Detector is a Streamlit application for classifying chest X-rays as Normal or Viral Pneumonia. Users upload one or more images; the app validates and decodes them, runs ResNet-18, KNN, and SVM inference, averages their two-class probabilities, and presents predictions, confidence, charts, and downloadable batch results. The README documents the major capabilities; the sampled app.py excerpts show validation, classical-model preprocessing, inference calls, and result rendering, but do not expose every internal implementation detail.

Main entry points: app.py Upload and UI · resnet18_state_dict.pth ResNet-18 model · knn_model_latest.pkl KNN bundle · svm_rbf_latest.pkl SVM bundle

Components 12

Upload interface

Upload and UI app.py

Inference workflow

Classifier models

ResNet-18 model ResNet18/resnet18_state_dict.pth
KNN bundle KNN/knn_model_latest.pkl
SVM bundle SVM/svm_rbf_latest.pkl

Prediction results

Other components

Image uploader

Connections 14

Image uploader → Upload and UI: uploads images README.md
Upload and UI → Image validation: validates uploads README.md
Image validation → Image preprocessing: decodes images app.py
Image preprocessing → ResNet-18 model: prepares input README.md
Image preprocessing → KNN bundle: prepares features app.py
Image preprocessing → SVM bundle: prepares features app.py
Upload and UI → ResNet-18 model: loads and runs README.md
Upload and UI → KNN bundle: loads and runs README.md
Upload and UI → SVM bundle: loads and runs README.md
Upload and UI → Probability ensemble: combines scores README.md
Probability ensemble → Prediction display: selects prediction app.py
Prediction display → Confidence chart: shows confidence app.py
Prediction display → Batch summary CSV: exports summary README.md
Prediction display → Session history: updates session app.py

Drawn by GitDiagram on November 14, 2026 from the repository's file tree, README and source files. Read as Markdown ·