Update app.py

app.py

@@ -8,6 +8,7 @@ import subprocess
 import numpy as np
 import os
 from threading import Thread
+import io
 
 # Model and Processor Loading (Done once at startup)
 MODEL_ID = "Qwen/Qwen2-VL-7B-Instruct"
@@ -24,17 +25,38 @@ image_extensions = Image.registered_extensions()
 video_extensions = ("avi", "mp4", "mov", "mkv", "flv", "wmv", "mjpeg", "wav", "gif", "webm", "m4v", "3gp")
 
 
-@spaces.GPU
-def qwen_inference(media_path, text_input=None):
-    print(media_path)
-    if media_path.endswith(tuple([i for i, f in image_extensions.items()])):
+def identify_and_save_blob(blob):
+    """Identifies if the blob is an image or video and saves it accordingly."""
+    try:
+        Image.open(io.BytesIO(blob)).verify()  # Check if it's a valid image
+        extension = ".png"
         media_type = "image"
-    elif media_path.endswith(video_extensions):  # Check if it's a video path
+    except:
+        extension = ".mp4"
         media_type = "video"
-    else:
-        raise ValueError(
-            "Unsupported media type. Please upload an image or video."
-        )
+
+    filename = f"temp_media{extension}"
+    with open(filename, "wb") as f:
+        f.write(blob)
+    return filename, media_type
+
+
+@spaces.GPU
+def qwen_inference(media_input, text_input=None):
+    if isinstance(media_input, str):  # If it's a filepath
+        media_path = media_input
+        if media_path.endswith(tuple([i for i, f in image_extensions.items()])):
+            media_type = "image"
+        elif media_path.endswith(video_extensions): 
+            media_type = "video"
+        else:
+            raise ValueError(
+                "Unsupported media type. Please upload an image or video."
+            )
+    else:  # If it's a blob
+        media_path, media_type = identify_and_save_blob(media_input)
+
+    print(media_path)
 
     messages = [
         {
@@ -75,7 +97,6 @@ def qwen_inference(media_path, text_input=None):
         buffer += new_text
         yield buffer
 
-
 css = """
   #output {
     height: 500px; 
@@ -91,7 +112,7 @@ with gr.Blocks(css=css) as demo:
         with gr.Row():
             with gr.Column():
                 input_media = gr.File(
-                    label="Upload Image or Video", type="filepath"
+                    label="Upload Image or Video", type="filepath" 
                 )
                 text_input = gr.Textbox(label="Question")
                 submit_btn = gr.Button(value="Submit")