Update pages/1_Clothing_Bias.py

pages/1_Clothing_Bias.py

@@ -8,32 +8,27 @@ from transformers import CLIPProcessor, CLIPModel
 import pandas as pd
 import matplotlib.pyplot as plt
 
-# Set page config
-st.set_page_config(page_title="Clothing Bias in Scene Classification", layout="wide")
-
+# Load model
 @st.cache_resource
 def load_clip_model():
     model = CLIPModel.from_pretrained("openai/clip-vit-base-patch32")
     processor = CLIPProcessor.from_pretrained("openai/clip-vit-base-patch32", use_fast=False)
     return model, processor
 
-st.header("Clothing Bias in Scene Classification")
+# --- Main page content ---
+st.header("👖 Clothing Bias in Scene Classification 👗")
 st.markdown("""
 This application explores biases in scene classification models related to clothing attributes.
 It leverages the CLIP model to analyze and highlight these biases.
 """)
 
-# Image input
 input_method = st.selectbox("Select Input Method", ["Default Images", "Upload Image", "Use Image URL"], index=0)
 
 image = None
-default_dir = "default_images/clothing_bias"
-
 if input_method == "Upload Image":
     uploaded_file = st.file_uploader("Upload your own image", type=["jpg", "png", "jpeg"])
     if uploaded_file:
         image = Image.open(uploaded_file).convert("RGB")
-
 elif input_method == "Use Image URL":
     image_url = st.text_input("Paste an image URL")
     if image_url:
@@ -42,30 +37,39 @@ elif input_method == "Use Image URL":
             image = Image.open(BytesIO(response.content)).convert("RGB")
         except:
             st.error("Couldn't load image from the provided URL.")
-
 elif input_method == "Default Images":
-    default_images = sorted([f for f in os.listdir(default_dir) if f.lower().endswith((".jpg", ".png", ".jpeg"))])
-    cols = st.columns(4)
-    selected_file = None
+    st.subheader("🖼️ Select a Default Image")
 
-    for i, img_name in enumerate(default_images):
-        col = cols[i % 4]
-        img_path = os.path.join(default_dir, img_name)
+    image_dir = "default_images/clothing_bias"
+    default_images = sorted([f for f in os.listdir(image_dir) if f.lower().endswith((".jpg", ".jpeg", ".png"))])
+
+    selected_image = None
+    columns = st.columns(4)  # Display images in 4 columns
+
+    for i, image_file in enumerate(default_images):
+        col = columns[i % 4]
+        img_path = os.path.join(image_dir, image_file)
         with col:
-            st.image(img_path, caption=chr(65 + i), use_container_width=True)
-            if st.button(f"Select {chr(65 + i)}", key=f"select_{i}"):
-                selected_file = img_path
+            st.image(img_path, caption=image_file, use_column_width=True)
+            if st.button(f"Select {image_file}", key=image_file):
+                selected_image = image_file
+
+    # Store selected image using session state so selection persists
+    if selected_image:
+        st.session_state.selected_image = selected_image
 
-    if selected_file:
-        image = Image.open(selected_file).convert("RGB")
+    if "selected_image" in st.session_state:
+        image_path = os.path.join(image_dir, st.session_state.selected_image)
+        image = Image.open(image_path).convert("RGB")
+        st.success(f"Selected: {st.session_state.selected_image}")
 
-# Display the selected image
+# Show the image if loaded
 if image is not None:
-    st.image(image, caption="Input Image", use_container_width=True)
+    st.image(image, caption="Input Image", width=250)
 
 # Prompt input
-st.subheader("Candidate Scene Labels")
-default_prompts = ["a business executive", "a festival participant", "an Indian maid", "a school teacher"]
+st.subheader("📝 Candidate Scene Labels")
+default_prompts = ["a business executive", "a festival participant"]
 prompts_text = st.text_area("Enter one label per line:", "\n".join(default_prompts))
 labels = [label.strip() for label in prompts_text.strip().split("\n") if label.strip()]
 
@@ -81,13 +85,11 @@ if st.button("🔍 Analyze Image"):
                 outputs = model(**inputs)
                 probs = outputs.logits_per_image.softmax(dim=1)[0]
 
-        # Display classification
+        # Show probabilities
         st.subheader("📊 Classification Probabilities")
-        df = pd.DataFrame({
-            "Label": labels,
-            "Probability": probs.numpy()
-        })
-        df.index += 1
+        data = {"Label": labels, "Probability": probs.numpy()}
+        df = pd.DataFrame(data)
+        df.index += 1  # Start index from 1
         st.table(df)
         st.write("**Most likely label**:", labels[probs.argmax().item()])
         st.write("\n")