Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -2,9 +2,6 @@ import streamlit as st
|
|
2 |
import torch
|
3 |
from transformers import AutoProcessor, PaliGemmaForConditionalGeneration
|
4 |
from PIL import Image
|
5 |
-
from transformers import AutoProcessor, AutoModelForImageTextToText
|
6 |
-
|
7 |
-
|
8 |
|
9 |
# Load the model and processor
|
10 |
model_id = "brucewayne0459/paligemma_derm"
|
@@ -38,7 +35,7 @@ st.markdown(
|
|
38 |
|
39 |
# Streamlit app title and instructions
|
40 |
st.title("Skin Condition Identifier")
|
41 |
-
st.write("Upload an image and provide a
|
42 |
|
43 |
# Column layout for input and display
|
44 |
col1, col2 = st.columns([3, 2])
|
@@ -46,51 +43,41 @@ col1, col2 = st.columns([3, 2])
|
|
46 |
with col1:
|
47 |
# File uploader for image
|
48 |
uploaded_file = st.file_uploader("Choose an image...", type=["jpg", "jpeg", "png"])
|
49 |
-
prompt = "Identify the skin condition?"
|
50 |
-
|
51 |
-
# Text input for prompt
|
52 |
-
input_text = st.text_input("Enter your prompt:", prompt)
|
53 |
-
|
54 |
with col2:
|
55 |
-
# Display uploaded image (if any)
|
56 |
if uploaded_file:
|
|
|
57 |
input_image = Image.open(uploaded_file).convert("RGB")
|
58 |
-
|
59 |
-
# Resize image for display (300x300 pixels)
|
60 |
resized_image = input_image.resize((300, 300))
|
61 |
-
|
62 |
-
# Display the resized image
|
63 |
st.image(resized_image, caption="Uploaded Image (300x300)", use_container_width=True)
|
64 |
|
65 |
-
|
66 |
-
|
67 |
-
|
68 |
-
|
69 |
-
|
|
|
70 |
try:
|
71 |
-
# Resize image for processing (512x512 pixels)
|
72 |
-
max_size = (512, 512)
|
73 |
-
input_image = input_image.resize(max_size)
|
74 |
-
|
75 |
# Prepare inputs
|
76 |
-
|
77 |
-
|
78 |
-
|
79 |
-
|
80 |
-
|
81 |
-
|
82 |
-
|
83 |
-
|
84 |
-
|
85 |
-
|
86 |
-
|
87 |
-
|
88 |
-
|
89 |
-
|
90 |
-
|
91 |
-
|
92 |
# Display result
|
93 |
st.success("Analysis Complete!")
|
94 |
st.write("**Model Output:**", decoded_output)
|
|
|
95 |
except Exception as e:
|
96 |
st.error(f"Error: {str(e)}")
|
|
|
2 |
import torch
|
3 |
from transformers import AutoProcessor, PaliGemmaForConditionalGeneration
|
4 |
from PIL import Image
|
|
|
|
|
|
|
5 |
|
6 |
# Load the model and processor
|
7 |
model_id = "brucewayne0459/paligemma_derm"
|
|
|
35 |
|
36 |
# Streamlit app title and instructions
|
37 |
st.title("Skin Condition Identifier")
|
38 |
+
st.write("Upload an image and provide a custom prompt to identify the skin condition.")
|
39 |
|
40 |
# Column layout for input and display
|
41 |
col1, col2 = st.columns([3, 2])
|
|
|
43 |
with col1:
|
44 |
# File uploader for image
|
45 |
uploaded_file = st.file_uploader("Choose an image...", type=["jpg", "jpeg", "png"])
|
46 |
+
prompt = st.text_input("Enter your prompt:", "Identify the skin condition?")
|
47 |
+
|
|
|
|
|
|
|
48 |
with col2:
|
|
|
49 |
if uploaded_file:
|
50 |
+
# Open and resize the uploaded image
|
51 |
input_image = Image.open(uploaded_file).convert("RGB")
|
|
|
|
|
52 |
resized_image = input_image.resize((300, 300))
|
|
|
|
|
53 |
st.image(resized_image, caption="Uploaded Image (300x300)", use_container_width=True)
|
54 |
|
55 |
+
# Resize image for processing (512x512 pixels)
|
56 |
+
max_size = (512, 512)
|
57 |
+
processed_image = input_image.resize(max_size)
|
58 |
+
|
59 |
+
# Predict automatically when the image is uploaded or the prompt changes
|
60 |
+
with st.spinner("Processing..."):
|
61 |
try:
|
|
|
|
|
|
|
|
|
62 |
# Prepare inputs
|
63 |
+
inputs = processor(
|
64 |
+
text=prompt,
|
65 |
+
images=processed_image,
|
66 |
+
return_tensors="pt",
|
67 |
+
padding="longest"
|
68 |
+
).to(device)
|
69 |
+
|
70 |
+
# Generate output
|
71 |
+
default_max_tokens = 50 # Set a default value for max tokens
|
72 |
+
with torch.no_grad():
|
73 |
+
outputs = model.generate(**inputs, max_new_tokens=default_max_tokens)
|
74 |
+
|
75 |
+
# Decode output
|
76 |
+
decoded_output = processor.decode(outputs[0], skip_special_tokens=True)
|
77 |
+
|
|
|
78 |
# Display result
|
79 |
st.success("Analysis Complete!")
|
80 |
st.write("**Model Output:**", decoded_output)
|
81 |
+
|
82 |
except Exception as e:
|
83 |
st.error(f"Error: {str(e)}")
|