replace Qwen2-VL with Florence-2 API for caption generation

app.py

@@ -87,7 +87,7 @@ def generate_captions_api(image):
                 "max_tokens": 80
             }
             response = requests.post(
-                QWEN_VL_URL,
+                
                 headers=HF_HEADERS,
                 json=payload,
                 timeout=40
@@ -100,7 +100,7 @@ def generate_captions_api(image):
                 st.warning("Qwen2-VL API error: " + str(response.status_code) + " " + response.text[:100])
                 captions.append("a scene with various objects and people")
         except Exception as e:
-            st.warning("Qwen2-VL exception: " + str(e))
+            st.warning("Florence-2 exception: " + str(e))
             captions.append("a scene captured in the image")
 
     seen, unique = set(), []