Comments and unused variables removed

examples/efficientdet/debugger.py

@@ -1,14 +1,14 @@
-import tensorflow as tf
-# gpus = tf.config.experimental.list_physical_devices('GPU')
-# tf.config.experimental.set_memory_growth(gpus[0], True)
-
-
 import numpy as np
-from efficientdet import EFFICIENTDETD0
-from paz.datasets import VOC
-from paz.abstract import Processor, SequentialProcessor
+import tensorflow as tf
 from paz import processors as pr
+from paz.abstract import Processor, SequentialProcessor
+from paz.datasets import VOC
+
 from detection import AugmentDetection
+from efficientdet import EFFICIENTDETD0
+
+# gpus = tf.config.experimental.list_physical_devices('GPU')
+# tf.config.experimental.set_memory_growth(gpus[0], True)
 
 
 class ShowBoxes(Processor):

examples/efficientdet/detection.py

@@ -107,6 +107,17 @@ def call(self, image):
         return self.wrap(image, outputs)
 
 
+class DetectSingleShot(DetectSingleShot):
+    def __init__(
+            self, model, class_names, score_thresh, nms_thresh,
+            mean=pr.BGR_IMAGENET_MEAN, variances=[0.1, 0.1, 0.2, 0.2],
+            draw=True):
+        super().__init__(
+            model, class_names, score_thresh, nms_thresh,
+            mean, variances, draw)
+        self.draw_boxes2D = DrawBoxes2D(class_names)
+
+
 class DrawBoxes2D(pr.DrawBoxes2D):
     def __init__(
             self, class_names=None, colors=None,
@@ -131,14 +142,12 @@ def call(self, image, boxes2D):
         for box2D in boxes2D:
             prediction_parameters = self.compute_prediction_parameters(box2D)
             x_min, y_min, x_max, y_max, color, text = prediction_parameters
-            draw_opaque_box(
-                image, (x_min, y_min), (x_max, y_max), color)
+            draw_opaque_box(image, (x_min, y_min), (x_max, y_max), color)
         image = make_box_transparent(raw_image, image)
         for box2D in boxes2D:
             prediction_parameters = self.compute_prediction_parameters(box2D)
             x_min, y_min, x_max, y_max, color, text = prediction_parameters
-            add_box_border(
-                image, (x_min, y_min), (x_max, y_max), color, 2)
+            add_box_border(image, (x_min, y_min), (x_max, y_max), color, 2)
             text_size = get_text_size(text, self.scale, 1)
             (text_W, text_H), _ = text_size
             draw_opaque_box(
@@ -147,14 +156,3 @@ def call(self, image, boxes2D):
             put_text(
                 image, text, (x_min+2, y_min + 17), self.scale, (0, 0, 0), 1)
         return image
-
-
-class DetectSingleShot(DetectSingleShot):
-    def __init__(
-            self, model, class_names, score_thresh, nms_thresh,
-            mean=pr.BGR_IMAGENET_MEAN, variances=[0.1, 0.1, 0.2, 0.2],
-            draw=True):
-        super().__init__(
-            model, class_names, score_thresh, nms_thresh,
-            mean, variances, draw)
-        self.draw_boxes2D = DrawBoxes2D(class_names)

examples/efficientdet/draw.py

@@ -1,9 +1,7 @@
 import cv2
 
-GREEN = (0, 255, 0)
 FONT = cv2.FONT_HERSHEY_SIMPLEX
 LINE = cv2.LINE_AA
-FILLED = cv2.FILLED
 
 
 def put_text(image, text, point, scale, color, thickness):
@@ -20,7 +18,6 @@ def put_text(image, text, point, scale, color, thickness):
     # Returns
         Numpy array with shape ``[H, W, 3]``. Image with text.
     """
-    # cv2.putText returns an image in contrast to other drawing cv2 functions.
     return cv2.putText(image, text, point, FONT, scale, color, thickness, LINE)
 
 

examples/efficientdet/efficientdet_test.py

@@ -203,50 +203,3 @@ def test_efficientdet_result(model_input_output_base_path, model,
         assert np.all(model()(preprocessed_input).numpy() ==
                       target_model_output), 'Model result not as expected'
     del model
-
-# def test_feature_fusion_sum():
-#     nodes1 = tf.constant([1, 3])
-#     nodes2 = tf.constant([1, 3])
-#     feature_node = FeatureNode(6, [3, 4], 3, True, True, True, False, 'sum',
-#                                None)
-#     output_node = feature_node.fuse_features([nodes1, nodes2])
-#     expected_node = tf.constant([2, 6], dtype=tf.int32)
-#     check_equality = tf.math.equal(output_node, expected_node)
-#     check_flag = tf.reduce_all(check_equality)
-#     assert check_flag, 'Feature fusion - \'sum\' mismatch'
-
-
-# def test_feature_fusion_attention():
-#     nodes1 = tf.constant([1, 3], dtype=tf.float32)
-#     nodes2 = tf.constant([1, 3], dtype=tf.float32)
-#     feature_node = FeatureNode(6, [3, 4], 3, True, True, True, False,
-#                                'attention', None)
-#     feature_node.build((10, 128, 128, 3))
-#     output_node = feature_node.fuse_features([nodes1, nodes2])
-#     expected_node = tf.constant([1.0, 3.0], dtype=tf.float32)
-#     check_equality = tf.math.equal(output_node, expected_node)
-#     check_flag = tf.reduce_all(check_equality)
-#     assert check_flag, 'Feature fusion - attention method mismatch'
-
-
-# def test_feature_fusion_fastattention():
-#     nodes1 = tf.constant([1, 3], dtype=tf.float32)
-#     nodes2 = tf.constant([1, 3], dtype=tf.float32)
-#     feature_node = FeatureNode(6, [3, 4], 3, True, True, True, False,
-#                                'fastattention', None)
-#     feature_node.build((10, 128, 128, 3))
-#     output_node = feature_node.fuse_features([nodes1, nodes2])
-#     expected_node = tf.constant([0.99995005, 2.9998503], dtype=tf.float32)
-#     check_equality = tf.math.equal(output_node, expected_node)
-#     check_flag = tf.reduce_all(check_equality)
-#     assert check_flag, 'Feature fusion - fastattention method mismatch'
-
-
-# test_efficientdet_model()
-# test_efficientnet_model()
-# test_efficientnet_bottleneck_block()
-# test_efficientnet_se_block()
-# test_all_efficientdet_models(get_model_path)
-# test_feature_fusion_sum()
-# test_feature_fusion_attention()
-# test_feature_fusion_fastattention()

examples/efficientdet/evaluate_mAP.py

@@ -1,10 +1,11 @@
+import os
+
+import tensorflow as tf
 from paz.datasets import VOC
 from paz.optimization.callbacks import EvaluateMAP
 from paz.pipelines import DetectSingleShot
-from efficientdet import EFFICIENTDETD0
-import tensorflow as tf
-import os
 
+from efficientdet import EFFICIENTDETD0
 
 if __name__ == "__main__":
     weights_path = "./trained_models/efficientdet-d0/"

examples/efficientdet/infer_efficientdet.py

@@ -5,7 +5,6 @@
 from efficientdet import EFFICIENTDETD0
 from utils import raw_images
 
-
 if __name__ == "__main__":
     model = EFFICIENTDETD0(num_classes=21, base_weights='COCO',
                            head_weights=None)