dedup images and add captions

app.py

@@ -15,13 +15,6 @@ professions_dset = load_from_disk("professions")
 professions_df = professions_dset.to_pandas()
 
 
-def get_image(model, fname):
-    return professions_dset.select(
-        professions_df[
-            (professions_df["image_path"] == fname) & (professions_df["model"] == model)
-        ].index
-    )["image"][0]
-
 
 clusters_dicts = dict(
     (num_cl, json.load(open(f"clusters/professions_to_clusters_{num_cl}.json")))
@@ -158,13 +151,28 @@ def make_profession_table(num_clusters, prof_names, mod_name, max_cols=8):
         .to_html()
     )
 
+def get_image(model, fname, score):
+    return (
+        professions_dset.select(
+            professions_df[
+                (professions_df["image_path"] == fname) & (professions_df["model"] == model)
+            ].index
+        )["image"][0],
+        " ".join(fname.split("/")[0].split("_")[4:]) + f" | {score:.2f}" + f" | {models[model]}"
+    )
+
 
-def show_examplars(num_clusters, prof_name, cl_id):
+def show_examplars(num_clusters, prof_name, cl_id, confidence_threshold=0.5):
+    # only show images where the similarity to the centroid is > 0.7
     examplars_dict = clusters_dicts[num_clusters]["All"][prof_name][
         "cluster_examplars"
     ][str(cl_id)]
-    l = list(chain(*[examplars_dict[k] for k in examplars_dict]))
-    return [get_image(model, fname) for _, model, fname in l]
+    l = [tuple(img) for img in examplars_dict["close"] + examplars_dict["mid"][:2] + examplars_dict["far"]]
+    l = [img for i, img in enumerate(l) if img[0] > confidence_threshold and img not in l[:i]]
+    return (
+        [get_image(model, fname, score) for score, model, fname in l],
+        gr.update(label=f"Generations for profession ''{prof_name}'' assigned to cluster {cl_id} of {num_clusters}")
+    )
 
 
 with gr.Blocks(title=TITLE) as demo:
@@ -261,7 +269,7 @@ with gr.Blocks(title=TITLE) as demo:
         with gr.Row():
             examplars_plot = gr.Gallery(
                 label="Profession images assigned to the selected cluster."
-            ).style(grid=5, height="auto")
+            ).style(grid=4, height="auto", container=True)
             demo.load(
                 show_examplars,
                 [
@@ -269,10 +277,10 @@ with gr.Blocks(title=TITLE) as demo:
                     profession_choice_focus,
                     cluster_id_focus,
                 ],
-                examplars_plot,
+                [examplars_plot, examplars_plot],
                 queue=False,
             )
-            for var in [cluster_id_focus]:
+            for var in [num_clusters_focus, profession_choice_focus, cluster_id_focus]:
                 var.change(
                     show_examplars,
                     [
@@ -280,7 +288,7 @@ with gr.Blocks(title=TITLE) as demo:
                         profession_choice_focus,
                         cluster_id_focus,
                     ],
-                    examplars_plot,
+                    [examplars_plot, examplars_plot],
                     queue=False,
                 )