README.md

metadata

title: Pneumonia CNN Inference
emoji: 🩺
colorFrom: blue
colorTo: indigo
sdk: gradio
sdk_version: 4.44.0
app_file: app.py
pinned: false
python_version: 3.1
short_description: Classify chest X-ray images as Pneumonia or Normal.
tags:
  - medical-imaging
  - pneumonia-detection
  - tensorflow
  - gradio

Pneumonia CNN Inference Space

This Hugging Face Space uses a Keras CNN to classify chest X-ray images as Pneumonia or Normal.

How to Use

1. Upload a chest X-ray image (JPEG) using the Gradio interface.
2. View the prediction (PNEUMONIA or NORMAL) and probability.

Model Details

• Dataset: chest-xray-pneumonia
• Test Accuracy: ~90.54%
• Model Format: TensorFlow SavedModel
• Classification Report:

                     precision    recall  f1-score   support

Pneumonia (Class 0)       0.95      0.89      0.92       390
   Normal (Class 1)       0.84      0.92      0.88       234

           accuracy                           0.91       624
          macro avg       0.90      0.91      0.90       624
       weighted avg       0.91      0.91      0.91       624

Local Inference

import tensorflow as tf
import numpy as np
import cv2

model = tf.saved_model.load("pneumonia_cnn_saved_model")
infer = model.signatures['serving_default']
class_names = ['PNEUMONIA', 'NORMAL']

def preprocess_image(image_path):
    img = cv2.imread(image_path, cv2.IMREAD_GRAYSCALE)
    img = cv2.resize(img, (150, 150)) / 255.0
    img = img.reshape(1, 150, 150, 1).astype(np.float32)
    return img

image = preprocess_image("path/to/xray.jpg")
prediction = infer(tf.convert_to_tensor(image))['dense_1'].numpy()
class_id = (prediction > 0.5).astype("int32")[0][0]
print("Prediction: ", class_names[class_id], " Probability: "prediction[0][0]:.4f)