feat: switch VQA model

app.py

@@ -1,25 +1,32 @@
+import torch
 import gradio as gr
-from transformers import pipeline
+from transformers import AutoModel, pipeline, AutoTokenizer
 
-pipeline = pipeline(task="image-classification", model="julien-c/hotdog-not-hotdog")
+inference = pipeline(task="visual-question-answering")
 
 
-def predict(input_img):
-    predictions = pipeline(input_img)
-    return input_img, {p["label"]: p["score"] for p in predictions}
+def predict(input_img, questions):
+    try:
+        predictions = inference(question=questions, image=input_img)
+        return str(predictions[0])
+    except Exception as e:
+        # 捕获异常，并将错误信息转换为字符串
+        error_message = str(e)
+        # 抛出gradio.Error来展示错误弹窗
+        raise gr.Error(error_message, duration=25)
 
 
 gradio_app = gr.Interface(
     predict,
-    inputs=gr.Image(
-        label="Select hot dog candidate", sources=["upload", "webcam"], type="pil"
-    ),
-    outputs=[
-        gr.Image(label="Processed Image"),
-        gr.Label(label="Result", num_top_classes=2),
+    inputs=[
+        gr.Image(
+            label="Select A Image", sources=["upload", "webcam"], type="pil"
+        ),
+        "text",
     ],
-    title="Hot Dog? Or Not?",
+    outputs="text",
+    title="Plz ask my anything",
 )
 
 if __name__ == "__main__":
-    gradio_app.launch()
+    gradio_app.launch(show_error=True, debug=True)