Narayana02 commited on
Commit
19e494c
Β·
verified Β·
1 Parent(s): 7ec2b4b

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +34 -29
app.py CHANGED
@@ -2,41 +2,48 @@ import streamlit as st
2
  from huggingface_hub import InferenceClient
3
  from config import HUGGINGFACE_API_KEY # Import your API key from a separate config file
4
  from PIL import Image
 
5
  from io import BytesIO
6
 
7
  # Streamlit App Configuration
8
  st.set_page_config(page_title="Llama-3.2 Demo App", page_icon="πŸ€–", layout="wide")
9
  st.title("πŸ–ΌοΈ Llama-3.2-90B-Vision-Instruct Demo App")
10
- st.markdown("<p style='text-align: center; font-size: 18px; color: #555;'>Upload an image and receive a text description of its content</p>", unsafe_allow_html=True)
11
 
12
- # User Inputs
13
- uploaded_image = st.file_uploader("Upload an Image", type=["png", "jpg", "jpeg"])
14
  user_prompt = st.text_input("Enter your prompt", value="Describe this image in a paragraph", placeholder="e.g., What is shown in the image?")
15
 
16
- # Function to display the uploaded image
17
- def show_uploaded_image(uploaded_image):
18
  try:
19
- img = Image.open(uploaded_image)
20
- st.image(img, caption="Uploaded Image", use_container_width=True)
21
- return img
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
22
  except Exception as e:
23
- st.error(f"❌ Unable to display image. Error: {e}")
24
- return None
25
 
26
  # Process user input
27
  if st.button("Get Description", key="get_description"):
28
- if uploaded_image and user_prompt:
29
  try:
30
- # Display the uploaded image
31
- img = show_uploaded_image(uploaded_image)
32
- if img is None:
33
- st.error("❌ Image processing failed.")
34
- st.stop()
35
-
36
- # Convert the image to bytes for model input
37
- img_buffer = BytesIO()
38
- img.save(img_buffer, format="PNG")
39
- img_bytes = img_buffer.getvalue()
40
 
41
  # Initialize the InferenceClient
42
  client = InferenceClient(api_key=HUGGINGFACE_API_KEY)
@@ -47,7 +54,7 @@ if st.button("Get Description", key="get_description"):
47
  "role": "user",
48
  "content": [
49
  {"type": "text", "text": user_prompt},
50
- {"type": "image", "image": {"bytes": img_bytes}}
51
  ]
52
  }
53
  ]
@@ -62,14 +69,16 @@ if st.button("Get Description", key="get_description"):
62
  # Extract JSON response
63
  model_response = completion.choices[0].message
64
 
65
- # Display the result
66
  st.subheader("πŸ“ Model Response")
 
 
67
  st.markdown(f"**Description**: {model_response.get('content', 'No description available')}")
68
 
69
  except Exception as e:
70
  st.error(f"❌ An error occurred: {e}")
71
  else:
72
- st.warning("⚠️ Please upload an image and enter a prompt.")
73
 
74
  # Clean UI Enhancements
75
  st.markdown("""
@@ -93,10 +102,6 @@ st.markdown("""
93
  border-radius: 10px;
94
  }
95
 
96
- .stFileUploader>div>div {
97
- border-radius: 10px;
98
- }
99
-
100
  /* Center the image */
101
  .stImage {
102
  display: block;
@@ -104,4 +109,4 @@ st.markdown("""
104
  margin-right: auto;
105
  }
106
  </style>
107
- """, unsafe_allow_html=True)
 
2
  from huggingface_hub import InferenceClient
3
  from config import HUGGINGFACE_API_KEY # Import your API key from a separate config file
4
  from PIL import Image
5
+ import requests
6
  from io import BytesIO
7
 
8
  # Streamlit App Configuration
9
  st.set_page_config(page_title="Llama-3.2 Demo App", page_icon="πŸ€–", layout="wide")
10
  st.title("πŸ–ΌοΈ Llama-3.2-90B-Vision-Instruct Demo App")
11
+ st.markdown("<p style='text-align: center; font-size: 18px; color: #555;'>Enter an image URL and get a description</p>", unsafe_allow_html=True)
12
 
13
+ # User Inputs with placeholder
14
+ image_url = st.text_input("Enter Image URL", value="", placeholder="Paste image URL here...", max_chars=400)
15
  user_prompt = st.text_input("Enter your prompt", value="Describe this image in a paragraph", placeholder="e.g., What is shown in the image?")
16
 
17
+ # Function to display the image from URL with height limit based on its actual size
18
+ def show_image_from_url(image_url, max_height=200):
19
  try:
20
+ response = requests.get(image_url)
21
+ img = Image.open(BytesIO(response.content))
22
+
23
+ # Get the original image size
24
+ img_width, img_height = img.size
25
+
26
+ # Calculate the new height and width based on the max height while maintaining the aspect ratio
27
+ if img_height > max_height:
28
+ aspect_ratio = img_width / img_height
29
+ new_height = max_height
30
+ new_width = int(new_height * aspect_ratio)
31
+ img_resized = img.resize((new_width, new_height))
32
+ else:
33
+ img_resized = img # No resizing needed if the image is smaller than the max height
34
+
35
+ # Center the image and display it
36
+ st.image(img_resized, caption=f"Source: {image_url}", use_container_width=True)
37
+
38
  except Exception as e:
39
+ st.error(f"❌ Unable to load image. Error: {e}")
 
40
 
41
  # Process user input
42
  if st.button("Get Description", key="get_description"):
43
+ if image_url and user_prompt:
44
  try:
45
+ # Show the image with dynamic resizing based on the image size
46
+ show_image_from_url(image_url, max_height=600)
 
 
 
 
 
 
 
 
47
 
48
  # Initialize the InferenceClient
49
  client = InferenceClient(api_key=HUGGINGFACE_API_KEY)
 
54
  "role": "user",
55
  "content": [
56
  {"type": "text", "text": user_prompt},
57
+ {"type": "image_url", "image_url": {"url": image_url}}
58
  ]
59
  }
60
  ]
 
69
  # Extract JSON response
70
  model_response = completion.choices[0].message
71
 
72
+ # Display the result in a clean and simple format
73
  st.subheader("πŸ“ Model Response")
74
+
75
+ # Display Content
76
  st.markdown(f"**Description**: {model_response.get('content', 'No description available')}")
77
 
78
  except Exception as e:
79
  st.error(f"❌ An error occurred: {e}")
80
  else:
81
+ st.warning("⚠️ Please enter an image URL and a prompt.")
82
 
83
  # Clean UI Enhancements
84
  st.markdown("""
 
102
  border-radius: 10px;
103
  }
104
 
 
 
 
 
105
  /* Center the image */
106
  .stImage {
107
  display: block;
 
109
  margin-right: auto;
110
  }
111
  </style>
112
+ """, unsafe_allow_html=True)