Update app.py

app.py

@@ -1,26 +1,41 @@
+import os
+import sys
 import json
+import logging
+import traceback
 import numpy as np
+import pandas as pd
 import gradio as gr
 import onnxruntime
 from PIL import Image
 from torchvision import transforms
-import pandas as pd
 
-# Load ONNX model
-ort_session = onnxruntime.InferenceSession("model_new_new_final.onnx")
+# Logging
+logging.basicConfig(level=logging.INFO, format="%(asctime)s %(levelname)s %(message)s", stream=sys.stdout)
+log = logging.getLogger("space")
 
-# Load metadata JSON
-with open('dat.json', 'r', encoding='utf-8') as f:
-    data = json.load(f)
+def log_exc(prefix):
+    etype, evalue, tb = sys.exc_info()
+    stack = "".join(traceback.format_exception(etype, evalue, tb))
+    log.error("%s: %s\n%s", prefix, evalue, stack)
+    return f"{prefix}: {evalue}"
 
-# Ensure deterministic class ordering
+# Load metadata
+with open("dat.json", "r", encoding="utf-8") as f:
+    data = json.load(f)
 keys = list(data.keys())
+log.info("Loaded %d classes from dat.json", len(keys))
+
+# Load ONNX
+ort = onnxruntime.InferenceSession("model_new_new_final.onnx")
+log.info("ONNX inputs: %s", [(i.name, i.shape, i.type) for i in ort.get_inputs()])
+log.info("ONNX outputs: %s", [(o.name, o.shape, o.type) for o in ort.get_outputs()])
 
-# Define transforms (PIL -> tensor NCHW)
-test_tfms = transforms.Compose([
+# Preprocess
+tfms = transforms.Compose([
     transforms.Resize((100, 100)),
     transforms.ToTensor(),
-    transforms.Normalize(mean=[0.7611, 0.5869, 0.5923], std=[0.1266, 0.1487, 0.1619])
+    transforms.Normalize(mean=[0.7611, 0.5869, 0.5923], std=[0.1266, 0.1487, 0.1619]),
 ])
 
 def probabilities_to_ints(probabilities, total_sum=100):
@@ -36,83 +51,61 @@ def probabilities_to_ints(probabilities, total_sum=100):
         rounded[int(np.argmax(probs))] += diff
     return rounded
 
-def Predict(image: Image.Image):
-    # Input is PIL.Image when inputs=gr.Image(type="pil")
-    img = image.convert("RGB")
-    tensor = test_tfms(img).unsqueeze(0).numpy().astype(np.float32)  # (1,C,H,W)
-
-    # ONNX inference
-    input_name = ort_session.get_inputs().name
-    outputs = ort_session.run(None, {input_name: tensor})
-    logits = outputs
-
-    # Flatten to 1D scores
-    if logits.ndim == 2:
-        scores = logits
-    elif logits.ndim == 1:
-        scores = logits
-    else:
-        raise ValueError(f"Unexpected logits shape: {logits.shape}")
-
-    # Predicted class
-    prediction_idx = int(np.argmax(scores))
-    disease_name = keys[prediction_idx]
-
-    # Lookup metadata (safe gets)
-    info = data.get(disease_name, {})
-    description = info.get('description', '')
-    symptoms = info.get('symptoms', '')
-    causes = info.get('causes', '')
-    treatment = info.get('treatment-1', info.get('treatment', ''))
-
-    # Build probabilities DataFrame for the bar plot
-    probs_int = probabilities_to_ints(scores)
-    df = pd.DataFrame({
-        "item": keys,
-        "probability": probs_int.astype(int)
-    })
-
-    # Return values matching declared outputs
-    return disease_name, description, symptoms, causes, treatment, df
-
-# Declare a BarPlot output component that will receive a DataFrame
-bar_output = gr.BarPlot(
-    x="item",
-    y="probability",
-    y_title="Probabilidad",
-    x_title="Nombre de la Enfermedad",
-    title="Distribucion de Probabilidad",
-    tooltip=["item", "probability"],
-    vertical=False
-)
+def predict(image: Image.Image):
+    try:
+        if image is None:
+            return "Error", "", "", "", "", pd.DataFrame({"item": keys, "probability": *len(keys)}), "No image provided"
+        pil = image.convert("RGB")
+        x = tfms(pil).unsqueeze(0).numpy().astype(np.float32)  # (1,C,H,W)
+        input_name = ort.get_inputs().name
+        outs = ort.run(None, {input_name: x})
+        logits = outs
+        if logits.ndim == 2:
+            scores = logits
+        elif logits.ndim == 1:
+            scores = logits
+        else:
+            raise ValueError(f"Unexpected logits shape: {logits.shape}")
+        if len(scores) != len(keys):
+            raise ValueError(f"Logits length {len(scores)} != classes {len(keys)}")
+        idx = int(np.argmax(scores))
+        name = keys[idx]
+        meta = data.get(name, {})
+        desc = meta.get("description", "")
+        symp = meta.get("symptoms", "")
+        causes = meta.get("causes", "")
+        treat = meta.get("treatment-1", meta.get("treatment", ""))
+        df = pd.DataFrame({"item": keys, "probability": probabilities_to_ints(scores).astype(int)})
+        return name, desc, symp, causes, treat, df, ""
+    except Exception:
+        err = log_exc("Inference failed")
+        df = pd.DataFrame({"item": keys if keys else ["N/A"], "probability": *(len(keys) if keys else 1)})
+        return "Error", "", "", "", "", df, err
 
-demo = gr.Interface(
-    fn=Predict,
-    inputs=gr.Image(type="pil", label="Imagen"),
-    outputs=[
-        gr.Textbox(label='Nombre de la Enfermedad'),
-        gr.Textbox(label='Descripcion'),
-        gr.Textbox(label='Sintomas'),
-        gr.Textbox(label='Causas'),
-        gr.Textbox(label='Tratamiento'),
-        bar_output
-    ],
-    title="Clasificacion de Enfermedades de la Piel",
-    description=(
-        'Este espacio se ha desarrollado como parte de una tesis para la Universidad Central de Venezuela '
-        'con el proposito de realizar diagnosticos precisos sobre una variedad de lesiones cutaneas. '
-        'Su objetivo es ayudar en la identificacion temprana y precisa de condiciones dermatologicas, incluyendo:\n\n'
-        '1) Queratosis Actinica \n\n'
-        '2) Carcinoma Basocelular \n\n'
-        '3) Dermatofibroma \n\n'
-        '4) Melanoma \n\n'
-        '5) Nevus \n\n'
-        '6) Queratosis Pigmentada Benigna \n\n'
-        '7) Queratosis Seborreica \n\n'
-        '8) Carcinoma de Celulas Escamosas \n\n'
-        '9) Lesion Vascular \n\n'
+with gr.Blocks(title="Clasificacion de Enfermedades de la Piel") as demo:
+    gr.Markdown("Suba una imagen y ejecute la prediccion.")
+    with gr.Row():
+        img = gr.Image(type="pil", label="Imagen")
+        with gr.Column():
+            out_name = gr.Textbox(label="Nombre de la Enfermedad")
+            out_desc = gr.Textbox(label="Descripcion")
+            out_symp = gr.Textbox(label="Sintomas")
+            out_causes = gr.Textbox(label="Causas")
+            out_treat = gr.Textbox(label="Tratamiento")
+    bar = gr.BarPlot(
+        x="item",
+        y="probability",
+        title="Distribucion de Probabilidad",
+        x_title="Nombre de la Enfermedad",
+        y_title="Probabilidad",
+        tooltip=["item", "probability"],
+        vertical=False,
+        label="Probabilidades"
     )
-)
+    err = gr.Textbox(label="Errores", interactive=False)
+    btn = gr.Button("Predecir")
+    btn.click(fn=predict, inputs=[img], outputs=[out_name, out_desc, out_symp, out_causes, out_treat, bar, err])
 
-# In Spaces, do not set share=True
-demo.launch(debug=True)
+if __name__ == "__main__":
+    # Spaces handles networking; no share=True
+    demo.launch(debug=True)