sync

src/streamlit_app.py

@@ -114,7 +114,13 @@ def information_conservation_score(x, x_hat, lambda1=0.5):
 
     return ((lambda1 * ms_ssim_score.mean() + (1 - lambda1) * freq_score.mean())).item()
 
-
+@st.cache_resource
+def load_my_dataset():
+    return load_dataset(
+        "rain-maker/RAW-RAIN-sample",
+        split="test",   # or use "imagefolder" if repo is raw files
+        repo_type="dataset"
+    )
 
 def get_image(scene, frame, kind):
     
@@ -126,19 +132,24 @@ def get_image(scene, frame, kind):
     )
     return Image.open(image)
 
+# def get_image(scene, frame, kind):
+#     # Construct relative path exactly like in your hf_hub_download version
+#     rel_path = f"{kind}_test/{scene}/rgb_output/output_{frame}.png"
+
+#     # Find the entry with that filename
+#     record = next((item for item in ds if item["image"].filename.endswith(rel_path)), None)
+
+#     if record is None:
+#         raise FileNotFoundError(f"{rel_path} not found in dataset")
+
+#     return record["image"]  # already a PIL.Image
+
 
 def validate_inputs(full_name):
     """Validate user inputs"""
     if not full_name.strip():
         return False, "Please enter your full name."
     
-    # if not user_id.strip():
-    #     return False, "Please enter your ID number."
-    
-    # Basic validation for ID (you can customize this)
-    # if not re.match(r'^[a-zA-Z0-9_-]+$', user_id.strip()):
-    #     return False, "ID number can only contain letters, numbers, hyphens, and underscores."
-    
     return True, ""
 
 def random_crop(gt_array, a_array, b_array, crop_size=512):
@@ -262,6 +273,7 @@ if not st.session_state.user_authenticated:
                 st.error(error_message)
 
 else:
+    ds = load_my_dataset()
     # Main image comparison interface
     st.header(f"Welcome, {st.session_state.full_name}")
     st.write(f"User ID: {st.session_state.user_id}")
@@ -313,9 +325,7 @@ else:
                     a_crop = st.session_state.current_crops['a']
                     b_crop = st.session_state.current_crops['b']
             
-            st.write(f"### Image set {st.session_state.index + 1}/{st.session_state.target_responses}")
-            # st.write(f"Scene: {st.session_state.current_scene}, Frame: {st.session_state.current_frame}")
-            # st.write("Compare Image A and Image B to the Ground Truth. Which one is closer to the Ground Truth?")
+            # st.write(f"### Image set {st.session_state.index + 1}/{st.session_state.target_responses}")
             
             # Display images
             cols = st.columns(3)
@@ -331,10 +341,6 @@ else:
                 cols[1].image(gt_crop, caption="Ground Truth", use_container_width=True)
                 cols[2].image(b_crop, caption="Image B", use_container_width=True)
             
-            # Choice selection
-            # choice = st.radio("Which image is closer to the Ground Truth?", 
-            #                  ["Image A (Left)", "Image B (Right)"],
-            #                  key=f"choice_{st.session_state.index}")
             
             col1, col2 = st.columns([1, 1])
             
@@ -476,35 +482,15 @@ else:
                         st.success("Choice and metrics recorded!")
                         st.rerun()
             
-            
-            # with col3:
-            #     if st.button("Reset Study"):
-            #         # Reset session state
-            #         for key in list(st.session_state.keys()):
-            #             del st.session_state[key]
-            #         st.rerun()
-            
-            # Progress and data display
+
             st.write(f"Completed comparisons: {st.session_state.index}")
-            
-            # Show collected responses
-            # if st.session_state.responses_data and st.checkbox("Show collected responses", value=False):
-            #     import pandas as pd
-            #     df = pd.DataFrame(st.session_state.responses_data, columns=st.session_state.csv_headers)
-            #     st.dataframe(df)
-            
-            # Display current metrics (optional, for debugging)
-            # if st.checkbox("Show current metrics", value=False):
-            #     with st.spinner("Computing current metrics..."):
-            #         current_metrics = compute_metrics(gt_crop, a_crop, b_crop, ics_lambda=0.5)
-            #         if current_metrics:
-            #             st.json(current_metrics)
+
             
     except Exception as e:
-        st.success("🎉 Study ended early!")
-        st.session_state.study_completed = True
-        st.rerun()
-        # st.error(f"Error loading images: {str(e)}")
-        # st.write("This might be due to network issues, the Hugging Face repository being unavailable, or missing dependencies.")
-        # st.write("Make sure you have the following packages installed:")
-        # st.code("pip install torch torchvision torchmetrics scikit-image opencv-python")
+        # st.success("🎉 Study ended early!")
+        # st.session_state.study_completed = True
+        # st.rerun()
+        st.error(f"Error loading images: {str(e)}")
+        st.write("This might be due to network issues, the Hugging Face repository being unavailable, or missing dependencies.")
+        st.write("Make sure you have the following packages installed:")
+        st.code("pip install torch torchvision torchmetrics scikit-image opencv-python")