From 8072eb526867cdec221827a6e7c93a034153bf2f Mon Sep 17 00:00:00 2001
From: digital-nomad-cheng <vincentcheng929@gmail.com>
Date: Fri, 13 Sep 2024 13:25:18 +0200
Subject: [PATCH 1/2] Add activation function option for detector.

---
 .../yolo_v8/yolo_v8_detector.py               | 27 ++++++++++---------
 1 file changed, 15 insertions(+), 12 deletions(-)

diff --git a/keras_cv/src/models/object_detection/yolo_v8/yolo_v8_detector.py b/keras_cv/src/models/object_detection/yolo_v8/yolo_v8_detector.py
index 8347e0fc7e..203bc7da39 100644
--- a/keras_cv/src/models/object_detection/yolo_v8/yolo_v8_detector.py
+++ b/keras_cv/src/models/object_detection/yolo_v8/yolo_v8_detector.py
@@ -104,7 +104,7 @@ def get_anchors(
     return all_anchors, all_strides
 
 
-def apply_path_aggregation_fpn(features, depth=3, name="fpn"):
+def apply_path_aggregation_fpn(features, depth=3, activation="swish", name="fpn"):
     """Applies the Feature Pyramid Network (FPN) to the outputs of a backbone.
 
     Args:
@@ -130,7 +130,7 @@ def apply_path_aggregation_fpn(features, depth=3, name="fpn"):
         channels=p4.shape[-1],
         depth=depth,
         shortcut=False,
-        activation="swish",
+        activation=activation,
         name=f"{name}_p4p5",
     )
 
@@ -142,7 +142,7 @@ def apply_path_aggregation_fpn(features, depth=3, name="fpn"):
         channels=p3.shape[-1],
         depth=depth,
         shortcut=False,
-        activation="swish",
+        activation=activation,
         name=f"{name}_p3p4p5",
     )
 
@@ -152,7 +152,7 @@ def apply_path_aggregation_fpn(features, depth=3, name="fpn"):
         p3p4p5.shape[-1],
         kernel_size=3,
         strides=2,
-        activation="swish",
+        activation=activation,
         name=f"{name}_p3p4p5_downsample1",
     )
     p3p4p5_d1 = ops.concatenate([p3p4p5_d1, p4p5], axis=-1)
@@ -160,7 +160,7 @@ def apply_path_aggregation_fpn(features, depth=3, name="fpn"):
         p3p4p5_d1,
         channels=p4p5.shape[-1],
         shortcut=False,
-        activation="swish",
+        activation=activation,
         name=f"{name}_p3p4p5_downsample1_block",
     )
 
@@ -171,7 +171,7 @@ def apply_path_aggregation_fpn(features, depth=3, name="fpn"):
         p3p4p5_d1.shape[-1],
         kernel_size=3,
         strides=2,
-        activation="swish",
+        activation=activation,
         name=f"{name}_p3p4p5_downsample2",
     )
     p3p4p5_d2 = ops.concatenate([p3p4p5_d2, p5], axis=-1)
@@ -179,7 +179,7 @@ def apply_path_aggregation_fpn(features, depth=3, name="fpn"):
         p3p4p5_d2,
         channels=p5.shape[-1],
         shortcut=False,
-        activation="swish",
+        activation=activation,
         name=f"{name}_p3p4p5_downsample2_block",
     )
 
@@ -189,6 +189,7 @@ def apply_path_aggregation_fpn(features, depth=3, name="fpn"):
 def apply_yolo_v8_head(
     inputs,
     num_classes,
+    activation="swish",
     name="yolo_v8_head",
 ):
     """Applies a YOLOV8 head.
@@ -229,14 +230,14 @@ def apply_yolo_v8_head(
             feature,
             box_channels,
             kernel_size=3,
-            activation="swish",
+            activation=activation,
             name=f"{cur_name}_box_1",
         )
         box_predictions = apply_conv_bn(
             box_predictions,
             box_channels,
             kernel_size=3,
-            activation="swish",
+            activation=activation,
             name=f"{cur_name}_box_2",
         )
         box_predictions = keras.layers.Conv2D(
@@ -249,14 +250,14 @@ def apply_yolo_v8_head(
             feature,
             class_channels,
             kernel_size=3,
-            activation="swish",
+            activation=activation,
             name=f"{cur_name}_class_1",
         )
         class_predictions = apply_conv_bn(
             class_predictions,
             class_channels,
             kernel_size=3,
-            activation="swish",
+            activation=activation,
             name=f"{cur_name}_class_2",
         )
         class_predictions = keras.layers.Conv2D(
@@ -400,6 +401,7 @@ def __init__(
         num_classes,
         bounding_box_format,
         fpn_depth=2,
+        activation="swish",
         label_encoder=None,
         prediction_decoder=None,
         **kwargs,
@@ -416,12 +418,13 @@ def __init__(
         features = list(feature_extractor(images).values())
 
         fpn_features = apply_path_aggregation_fpn(
-            features, depth=fpn_depth, name="pa_fpn"
+            features, depth=fpn_depth, activation=activation, name="pa_fpn"
         )
 
         outputs = apply_yolo_v8_head(
             fpn_features,
             num_classes,
+            activation=activation,
         )
 
         # To make loss metrics pretty, we use a no-op layer with a good name.

From 82f8e7fda71779a3ad74bfc7e27c9b8d0634d097 Mon Sep 17 00:00:00 2001
From: digital-nomad-cheng <vincentcheng929@gmail.com>
Date: Wed, 18 Sep 2024 17:51:54 +0200
Subject: [PATCH 2/2] Fix format.

---
 .../src/models/object_detection/yolo_v8/yolo_v8_detector.py   | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/keras_cv/src/models/object_detection/yolo_v8/yolo_v8_detector.py b/keras_cv/src/models/object_detection/yolo_v8/yolo_v8_detector.py
index 203bc7da39..6af20f01c7 100644
--- a/keras_cv/src/models/object_detection/yolo_v8/yolo_v8_detector.py
+++ b/keras_cv/src/models/object_detection/yolo_v8/yolo_v8_detector.py
@@ -104,7 +104,9 @@ def get_anchors(
     return all_anchors, all_strides
 
 
-def apply_path_aggregation_fpn(features, depth=3, activation="swish", name="fpn"):
+def apply_path_aggregation_fpn(
+    features, depth=3, activation="swish", name="fpn"
+):
     """Applies the Feature Pyramid Network (FPN) to the outputs of a backbone.
 
     Args: