Update app.py

app.py

@@ -124,13 +124,17 @@ def classify_zip_and_analyze_color(zip_file):
                 face_info = ""
                 try:
                     img_cv2 = cv2.cvtColor(np.array(image), cv2.COLOR_RGB2BGR)
-                    faces = DeepFace.analyze(img_cv2, actions=["age", "gender", "emotion"], enforce_detection=False)
-                    if isinstance(faces, list):  # multiple faces
+                    faces = DeepFace.analyze(
+                        img_cv2,
+                        actions=["age", "gender", "emotion"],
+                        enforce_detection=False
+                    )
+                    if isinstance(faces, list):
                         for f in faces:
                             face_info += f"Age: {f['age']}, Gender: {f['gender']}, Emotion: {f['dominant_emotion']}; "
-                    else:  # single face
+                    else:
                         face_info = f"Age: {faces['age']}, Gender: {faces['gender']}, Emotion: {faces['dominant_emotion']}"
-                except Exception as e:
+                except Exception:
                     face_info = "No face detected"
 
                 results.append((
@@ -181,32 +185,47 @@ def classify_zip_and_analyze_color(zip_file):
     plot2_img = Image.open(buf2)
 
     # ---------------------------
-    # Extract age and gender
+    # Extract age and weighted gender (handles gender dict)
     # ---------------------------
     ages = []
-    genders = []
+    gender_confidence = {"Man": 0, "Woman": 0}
 
     for info in df["Face Info"]:
         if info != "No face detected":
             for face_str in info.split(";"):
                 face_str = face_str.strip()
                 if face_str:
+                    # Age
                     age_part = face_str.split(",")[0]
                     age = int(age_part.replace("Age:", "").strip())
                     ages.append(age)
-                    
+
+                    # Gender parsing
                     gender_part = face_str.split(",")[1]
-                    gender = gender_part.replace("Gender:", "").strip()
-                    genders.append(gender)
+                    gender_dict_str = gender_part.replace("Gender:", "").strip()
+                    try:
+                        gender_dict = eval(gender_dict_str)
+                    except:
+                        continue
+
+                    # Take the highest probability as gender
+                    gender = max(gender_dict, key=gender_dict.get)
+                    conf = float(gender_dict[gender]) / 100  # convert % to 0-1
+
+                    # Weight capped at 0.9
+                    weight = min(conf, 0.9)
+                    if gender in gender_confidence:
+                        gender_confidence[gender] += weight
+                    else:
+                        gender_confidence[gender] = weight
 
     # ---------------------------
-    # Plot 3: Gender distribution
+    # Plot 3: Gender distribution (weighted ≤ 90%)
     # ---------------------------
     fig3, ax3 = plt.subplots()
-    gender_counts = pd.Series(genders).value_counts()
-    ax3.bar(gender_counts.index, gender_counts.values, color=["lightblue", "pink"])
-    ax3.set_title("Gender Distribution")
-    ax3.set_ylabel("Count")
+    ax3.bar(gender_confidence.keys(), gender_confidence.values(), color=["lightblue", "pink"])
+    ax3.set_title("Gender Distribution (Weighted ≤ 90% Confidence)")
+    ax3.set_ylabel("Sum of Confidence")
     buf3 = io.BytesIO()
     plt.savefig(buf3, format="png")
     plt.close(fig3)
@@ -240,7 +259,7 @@ demo = gr.Interface(
         gr.File(label="Download XLSX"),
         gr.Image(type="pil", label="Basic Color Frequency"),
         gr.Image(type="pil", label="Top Prediction Distribution"),
-        gr.Image(type="pil", label="Gender Distribution"),
+        gr.Image(type="pil", label="Gender Distribution (Weighted ≤90%)"),
         gr.Image(type="pil", label="Age Distribution"),
     ],
     title="Image Classifier with Color & Face Analysis",