Update app.py

app.py

@@ -1,44 +1,31 @@
-import streamlit as st
+import gradio as gr
 from PIL import Image
 import numpy as np
 import yolov5
 
-@st.cache_resource(show_spinner=False)
+@gr.cache(allow_output_mutation=True)
 def load_model():
     model = yolov5.load('keremberke/yolov5n-license-plate')
     model.conf = 0.25
     model.iou = 0.45
     return model
 
-def main():
-    st.title("🔍 Detector de Matrículas - YOLOv5")
-    st.write("Sube una imagen para detectar matrículas.")
-
-    img_file = st.file_uploader("Selecciona una imagen", type=["jpg", "jpeg", "png"])
-    if img_file is None:
-        st.info("Por favor sube una imagen.")
-        return
-
-    image = Image.open(img_file).convert("RGB")
-    st.image(image, caption="Imagen original", use_column_width=True)
-
+def detect(img: Image.Image):
+    img = img.convert("RGB")
     model = load_model()
-
-    img = np.array(image)
-
-    results = model(img, size=640)
-
-    results.render()  
-    st.image(results.ims[0], caption="Resultado con detecciones", use_column_width=True)
-
-   
-    detections = results.pred[0]
-    if detections is None or len(detections) == 0:
-        st.warning("No se detectaron matrículas.")
-    else:
-        st.success(f"Se detectaron {len(detections)} matrícula(s):")
-        for *box, conf, cls in detections.cpu().numpy():
-            st.write(f"📍 Confianza: {conf:.2f}")
+    arr = np.array(img)
+    results = model(arr, size=640)
+    results.render()
+    annotated = results.ims[0]
+    return Image.fromarray(annotated)
+
+demo = gr.Interface(
+    fn=detect,
+    inputs=gr.Image(type="pil"),
+    outputs=gr.Image(type="pil"),
+    title="🔍 Detector de Matrículas",
+    description="Sube una imagen y detecto matrículas usando YOLOv5n."
+)
 
 if __name__ == "__main__":
-    main()
+    demo.launch()