add image analytics

Author: rcmangasaryan

superannotate/__init__.py

@@ -8,6 +8,7 @@
 from .analytics.class_analytics import (
     attribute_distribution, class_distribution
 )
+from .analytics.user_analytics import image_analytics
 from .analytics.common import aggregate_annotations_as_df, df_to_annotations
 from .annotation_helpers import (
     add_annotation_bbox_to_json, add_annotation_comment_to_json,

superannotate/analytics/user_analytics.py

@@ -0,0 +1,56 @@
+import pandas as pd
+from tqdm import tqdm
+import plotly.express as px
+
+def image_analytics(annotations_df, visualize = False):
+    """
+    Aggregates image analytics: num instances/annotation time in seconds per image
+    :param annotations_df: pandas DataFrame of project annotations
+    :type annotations_df: pandas.DataFrame
+    :param visulaize: enables image analytics scatter plot
+    :type visualize: bool
+
+    :return: DataFrame on image analytics with columns ["image", "instances", "time"]
+    :rtype: pandas DataFrame
+
+    """
+    def fix_spent_time(grp: pd.Series) -> pd.Series:
+        grp = grp.copy()
+        grp_lost_msk = (grp > 600) | (grp.isna())
+        grp.loc[grp_lost_msk] = grp[~grp_lost_msk].median()
+        return grp
+    
+    analytics = {"user_id": [], "user_role": [], "image": [], "time": [], "ninstances": [] }
+    annot_cols = ["imageName", "instanceId", "createdAt", "creatorEmail", "creatorRole"]
+    annotations_df = annotations_df[annotations_df["creationType"] == "Manual"][annot_cols].drop_duplicates()
+
+    for annot, grp in annotations_df.groupby(["creatorEmail", "creatorRole"]):
+        grp_sorted = grp.sort_values("createdAt")
+        time_spent = grp_sorted.createdAt.diff().shift(-1).dt.total_seconds()
+        grp["time_spent"] = fix_spent_time(time_spent)
+        img_time = grp.groupby("imageName", as_index=False)["time_spent"].agg("sum")
+        img_n_instance = grp.groupby("imageName")["instanceId"].agg("count")
+
+        analytics["image"] += img_time.imageName.tolist()
+        analytics["time"] += img_time.time_spent.tolist()
+        analytics["ninstances"] += img_n_instance.tolist()
+        analytics["user_id"] += [annot[0]] * len(img_time)
+        analytics["user_role"] += [annot[1]] * len(img_time)
+
+    analytics_df = pd.DataFrame(analytics)
+    if visualize:
+        #scatter plot of number of instances vs annotation time
+        fig = px.scatter(
+            analytics_df,
+            x="ninstances",
+            y="time",
+            color="user_id",
+            facet_col="user_role",
+            custom_data = ["image"],
+            color_discrete_sequence=px.colors.qualitative.Dark24,
+            labels = {'user_id': "User Email", "ninstances": "Number of Instances", "time": "Annotation time"}
+        )
+        fig.for_each_annotation(lambda a: a.update(text=a.text.split("=")[-1]))
+        fig.update_traces(hovertemplate="%{customdata[0]}")
+        fig.show()
+    return analytics_df