Merge branch 'master' into panoptic-deeplab

official/vision/beta/ops/augment.py

@@ -1663,6 +1663,7 @@ class AutoAugment(ImageAugment):
 
     tf_policies = self._make_tf_policies()
     image, _ = select_and_apply_random_policy(tf_policies, image, bboxes=None)
+    image = tf.cast(image, dtype=input_image_type)
     return image
 
   def distort_with_boxes(self, image: tf.Tensor,

official/vision/beta/projects/centernet/configs/experiments/coco-centernet-hourglass-gpu.yaml

@@ -38,11 +38,11 @@ task:
   per_category_metrics: false
   weight_decay: 0.0005
   gradient_clip_norm: 10.0
-  annotation_file: 'coco/instances_val2017.json'
-  init_checkpoint: '/placer/prod/scratch/home/tf-model-garden-dev/vision/centernet/extremenet_hg104_512x512_coco17/2021-10-19'
+  annotation_file: '/readahead/200M/placer/prod/home/tensorflow-performance-data/datasets/coco/instances_val2017.json'
+  init_checkpoint: gs://tf_model_garden/vision/centernet/extremenet_hg104_512x512_coco17
   init_checkpoint_modules: 'backbone'
   train_data:
-    input_path: 'coco/train*'
+    input_path: '/readahead/200M/placer/prod/home/tensorflow-performance-data/datasets/coco/train*'
     drop_remainder: true
     dtype: 'float16'
     global_batch_size: 64
@@ -57,7 +57,7 @@ task:
       aug_rand_contrast: true
       odapi_augmentation: true
   validation_data:
-    input_path: 'coco/val*'
+    input_path: '/readahead/200M/placer/prod/home/tensorflow-performance-data/datasets/coco/val*'
     drop_remainder: false
     dtype: 'float16'
     global_batch_size: 16

official/vision/beta/projects/centernet/configs/experiments/coco-centernet-hourglass-tpu.yaml

@@ -37,11 +37,11 @@ task:
   per_category_metrics: false
   weight_decay: 0.0005
   gradient_clip_norm: 10.0
-  annotation_file: 'coco/instances_val2017.json'
-  init_checkpoint: '/placer/prod/scratch/home/tf-model-garden-dev/vision/centernet/extremenet_hg104_512x512_coco17/2021-10-19'
+  annotation_file: '/readahead/200M/placer/prod/home/tensorflow-performance-data/datasets/coco/instances_val2017.json'
+  init_checkpoint: gs://tf_model_garden/vision/centernet/extremenet_hg104_512x512_coco17
   init_checkpoint_modules: 'backbone'
   train_data:
-    input_path: 'coco/train*'
+    input_path: '/readahead/200M/placer/prod/home/tensorflow-performance-data/datasets/coco/train*'
     drop_remainder: true
     dtype: 'bfloat16'
     global_batch_size: 128
@@ -56,7 +56,7 @@ task:
       aug_rand_contrast: true
       odapi_augmentation: true
   validation_data:
-    input_path: 'coco/val*'
+    input_path: '/readahead/200M/placer/prod/home/tensorflow-performance-data/datasets/coco/val*'
     drop_remainder: false
     dtype: 'bfloat16'
     global_batch_size: 64

official/vision/beta/projects/panoptic_maskrcnn/configs/panoptic_maskrcnn.py

@@ -22,7 +22,7 @@ from official.core import config_definitions as cfg
 from official.core import exp_factory
 from official.modeling import hyperparams
 from official.modeling import optimization
-from official.vision.beta.projects.deepmac_maskrcnn.configs import deep_mask_head_rcnn as deepmac_maskrcnn
+from official.projects.deepmac_maskrcnn.configs import deep_mask_head_rcnn as deepmac_maskrcnn
 from official.vision.configs import common
 from official.vision.configs import maskrcnn
 from official.vision.configs import semantic_segmentation

official/vision/beta/projects/panoptic_maskrcnn/modeling/factory.py

@@ -16,9 +16,7 @@
 
 import tensorflow as tf
 
-
-from official.vision.beta.projects.panoptic_maskrcnn.configs import panoptic_deeplab as panoptic_deeplab_cfg
-from official.vision.beta.projects.deepmac_maskrcnn.tasks import deep_mask_head_rcnn
+from official.projects.deepmac_maskrcnn.tasks import deep_mask_head_rcnn
 from official.vision.beta.projects.panoptic_maskrcnn.configs import panoptic_maskrcnn as panoptic_maskrcnn_cfg
 from official.vision.beta.projects.panoptic_maskrcnn.modeling import panoptic_deeplab_model
 from official.vision.beta.projects.panoptic_maskrcnn.modeling.heads import panoptic_deeplab_heads

official/vision/beta/projects/panoptic_maskrcnn/modeling/panoptic_maskrcnn_model.py

@@ -18,7 +18,7 @@ from typing import List, Mapping, Optional, Union
 
 import tensorflow as tf
 
-from official.vision.beta.projects.deepmac_maskrcnn.modeling import maskrcnn_model
+from official.projects.deepmac_maskrcnn.modeling import maskrcnn_model
 
 
 class PanopticMaskRCNNModel(maskrcnn_model.DeepMaskRCNNModel):

official/vision/beta/projects/yolo/configs/yolo.py

@@ -187,6 +187,11 @@ class AnchorBoxes(hyperparams.Config):
   level_limits: Optional[List[int]] = None
   anchors_per_scale: int = 3
 
+  generate_anchors: bool = False
+  scaling_mode: str = 'sqrt'
+  box_generation_mode: str = 'per_level'
+  num_samples: int = 1024
+
   def get(self, min_level, max_level):
     """Distribute them in order to each level.
 
@@ -211,6 +216,9 @@ class AnchorBoxes(hyperparams.Config):
       start += self.anchors_per_scale
     return anchors_per_level, self.level_limits
 
+  def set_boxes(self, boxes):
+    self.boxes = [Box(box=box) for box in boxes]
+
 
 @dataclasses.dataclass
 class Yolo(hyperparams.Config):

official/vision/beta/projects/yolo/ops/kmeans_anchors_test.py

+# Copyright 2022 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""kmeans_test tests."""
+from absl.testing import parameterized
+import numpy as np
+import tensorflow as tf
+
+from official.vision.beta.projects.yolo.ops import kmeans_anchors
+
+
+class KMeansTest(parameterized.TestCase, tf.test.TestCase):
+
+  @parameterized.parameters((9, 3, 100))
+  def test_kmeans(self, k, anchors_per_scale, samples):
+    sample_list = []
+    for _ in range(samples):
+      boxes = tf.convert_to_tensor(np.random.uniform(0, 1, [k * 100, 4]))
+      sample_list.append({
+          "groundtruth_boxes": boxes,
+          "width": 10,
+          "height": 10
+      })
+
+    kmeans = kmeans_anchors.AnchorKMeans()
+    cl = kmeans(
+        sample_list, k, anchors_per_scale, image_resolution=[512, 512, 3])
+    cl = tf.convert_to_tensor(cl)
+    self.assertAllEqual(tf.shape(cl).numpy(), [k, 2])
+
+
+if __name__ == "__main__":
+  tf.test.main()

official/vision/beta/projects/yolo/ops/loss_utils.py

@@ -176,40 +176,43 @@ class GridGenerator:
     self._anchors = tf.convert_to_tensor(anchors)
     return
 
-  def _build_grid_points(self, lwidth, lheight, anchors, dtype):
+  def _build_grid_points(self, lheight, lwidth, anchors, dtype):
     """Generate a grid of fixed grid edges for box center decoding."""
     with tf.name_scope('center_grid'):
       y = tf.range(0, lheight)
       x = tf.range(0, lwidth)
-      num = tf.shape(anchors)[0]
       x_left = tf.tile(
-          tf.transpose(tf.expand_dims(y, axis=-1), perm=[1, 0]), [lwidth, 1])
-      y_left = tf.tile(tf.expand_dims(x, axis=-1), [1, lheight])
+          tf.transpose(tf.expand_dims(x, axis=-1), perm=[1, 0]), [lheight, 1])
+      y_left = tf.tile(tf.expand_dims(y, axis=-1), [1, lwidth])
       x_y = tf.stack([x_left, y_left], axis=-1)
       x_y = tf.cast(x_y, dtype=dtype)
+      num = tf.shape(anchors)[0]
       x_y = tf.expand_dims(
           tf.tile(tf.expand_dims(x_y, axis=-2), [1, 1, num, 1]), axis=0)
     return x_y
 
-  def _build_anchor_grid(self, anchors, dtype):
+  def _build_anchor_grid(self, height, width, anchors, dtype):
     """Get the transformed anchor boxes for each dimention."""
     with tf.name_scope('anchor_grid'):
       num = tf.shape(anchors)[0]
       anchors = tf.cast(anchors, dtype=dtype)
       anchors = tf.reshape(anchors, [1, 1, 1, num, 2])
+      anchors = tf.tile(anchors, [1, tf.cast(height, tf.int32),
+                                  tf.cast(width, tf.int32), 1, 1])
     return anchors
 
   def _extend_batch(self, grid, batch_size):
     return tf.tile(grid, [batch_size, 1, 1, 1, 1])
 
-  def __call__(self, width, height, batch_size, dtype=None):
+  def __call__(self, height, width, batch_size, dtype=None):
     if dtype is None:
       self.dtype = tf.keras.backend.floatx()
     else:
       self.dtype = dtype
-    grid_points = self._build_grid_points(width, height, self._anchors,
+    grid_points = self._build_grid_points(height, width, self._anchors,
                                           self.dtype)
     anchor_grid = self._build_anchor_grid(
+        height, width,
         tf.cast(self._anchors, self.dtype) /
         tf.cast(self._scale_anchors, self.dtype), self.dtype)
 

official/vision/ops/augment.py

@@ -1663,6 +1663,7 @@ class AutoAugment(ImageAugment):
 
     tf_policies = self._make_tf_policies()
     image, _ = select_and_apply_random_policy(tf_policies, image, bboxes=None)
+    image = tf.cast(image, dtype=input_image_type)
     return image
 
   def distort_with_boxes(self, image: tf.Tensor,
@@ -2259,7 +2260,7 @@ class MixupAndCutmix:
               labels: tf.Tensor) -> Tuple[tf.Tensor, tf.Tensor, tf.Tensor]:
     """Apply cutmix."""
     lam = MixupAndCutmix._sample_from_beta(self.cutmix_alpha, self.cutmix_alpha,
-                                           labels.shape)
+                                           tf.shape(labels))
 
     ratio = tf.math.sqrt(1 - lam)
 
@@ -2284,17 +2285,19 @@ class MixupAndCutmix:
         lambda x: _fill_rectangle(*x),
         (images, random_center_width, random_center_height, cut_width // 2,
          cut_height // 2, tf.reverse(images, [0])),
-        dtype=(tf.float32, tf.int32, tf.int32, tf.int32, tf.int32, tf.float32),
-        fn_output_signature=tf.TensorSpec(images.shape[1:], dtype=tf.float32))
+        dtype=(
+            images.dtype, tf.int32, tf.int32, tf.int32, tf.int32, images.dtype),
+        fn_output_signature=tf.TensorSpec(images.shape[1:], dtype=images.dtype))
 
     return images, labels, lam
 
   def _mixup(self, images: tf.Tensor,
              labels: tf.Tensor) -> Tuple[tf.Tensor, tf.Tensor, tf.Tensor]:
     lam = MixupAndCutmix._sample_from_beta(self.mixup_alpha, self.mixup_alpha,
-                                           labels.shape)
+                                           tf.shape(labels))
     lam = tf.reshape(lam, [-1, 1, 1, 1])
-    images = lam * images + (1. - lam) * tf.reverse(images, [0])
+    lam_cast = tf.cast(lam, dtype=images.dtype)
+    images = lam_cast * images + (1. - lam_cast) * tf.reverse(images, [0])
 
     return images, labels, tf.squeeze(lam)