Add classifier

app.py

@@ -1,7 +1,42 @@
+from fastai.vision.all import *
+from fastai.vision.core import PILImage
+import torch
 import gradio as gr
 
-def greet(name):
-    return "Hello " + name + "!!"
+import warnings
+warnings.filterwarnings("ignore")
 
-iface = gr.Interface(fn=greet, inputs="text", outputs="text")
-iface.launch()
+
+learn = load_learner('modelo.pkl')
+
+# Función de predicción
+def classify_image(image):
+    classes = ['abdomen', 'tobillo', 'col. cervical', 'torax', 'clavicula', 'codo', 'pies', 'dedos', 'antebrazo', 'mano', 'cadera',
+               'rodilla', 'pierna', 'col. lumbar', 'otros', 'pelvis', 'hombro', 'senos nasales', 'craneo', 'muslo', 'col. toracica', 'muñeca']
+
+    # Cargar imagen y realizar predicción
+    img = PILImage.create(image)
+    pred, pred_idx, probs = model.predict(img)
+
+    # Filtrar probabilidades
+    probs = torch.where(probs > 1e-2, probs, torch.tensor(0).to(probs.device))
+
+    # Obtener top 5 resultados
+    top5_probs, top5_idxs = torch.topk(probs, 5)
+    top5_classes = [classes[idx] for idx in top5_idxs]
+
+    # Crear lista de predicciones
+    predictions = []
+    for i in range(5):
+        if top5_probs[i] > 1e-2:
+            prob = round(float(top5_probs[i].numpy()), 3)
+            predictions.append(f"{top5_classes[i]}: {prob}")
+
+    return predictions
+
+
+inputs = gr.inputs.Image()
+outputs = gr.outputs.Textbox()
+
+gr.Interface(fn=classify_image, inputs=inputs, outputs=outputs, title='Clasificación de Imágenes Médicas',
+             description='Cargue una radiografía').launch()

modelo.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b07890782ac08210b5a6849a6bc73a4c3b4061e11c8885168f92e8a43184949
+size 114980453

requirements.txt

@@ -0,0 +1,7 @@
+gradio==3.34.0
+gradio_client==0.2.6
+torch==2.0.1
+torchvision==0.15.2
+timm==0.9.2
+fastai==2.7.12
+