Create app.py

app.py

@@ -0,0 +1,68 @@
+from itertools import combinations
+subgroups = []
+DEFAULT_RAW_DATA_PATH = "raw_data.json"
+with open(DEFAULT_RAW_DATA_PATH, 'r') as file:
+    raw_data = json.load(file)
+
+    
+for group in raw_data:
+    if len(group) > 3 and len(group)<6:
+        for size in [2, 3]:
+            subgroups.extend(list(combinations(group, size)))
+
+# Optional: include the original small groups (size <= 3)
+final_groups = [g for g in raw_data if len(g) <= 3] + [list(sg) for sg in subgroups]
+
+
+sample = random.sample(subgroups,100)
+sample_paths = []
+for wall in sample:
+    paths = [enc_id_to_path_drive(encoded_imgs_csv, int(enc_id)) for enc_id in wall]
+    sample_paths.append(paths)
+
+
+import gradio as gr
+from PIL import Image
+from pathlib import Path
+
+# ==== YOUR DATA ====
+# Each subgroup is a list of image paths
+
+results = []  # Store user feedback
+
+# ==== INTERFACE LOGIC ====
+def review_group(index, like):
+    if index < len(sample_paths):
+        decision = "like" if like else "dislike"
+        results.append({"group": sample_paths[index], "decision": decision})
+
+    next_index = index + 1
+    if next_index < len(sample_paths):
+        return (
+            [Image.open(p) for p in sample_paths[next_index]],
+            next_index
+        )
+    else:
+        return ([], next_index)
+
+# ==== UI ====
+with gr.Blocks() as demo:
+    group_index = gr.State(0)
+    image_gallery = gr.Gallery(label="Artwork Group", columns=3, height="auto")
+
+    with gr.Row():
+        like_btn = gr.Button("👍 Like")
+        dislike_btn = gr.Button("👎 Dislike")
+
+    # Actions
+    like_btn.click(review_group, [group_index, gr.State(True)], [image_gallery, group_index])
+    dislike_btn.click(review_group, [group_index, gr.State(False)], [image_gallery, group_index])
+
+    # Load first group
+    demo.load(fn=lambda: ([Image.open(p) for p in sample_paths[0]], 0), outputs=[image_gallery, group_index])
+
+# ==== Run ====
+demo.launch()
+
+results_df = pd.DataFrame(results)
+results_df.to_csv('results_df.csv')