Merge pull request #19 from srihari-humbarwadi/panoptic-deeplab-resnet50_fix

Use ResNet v1 as default backbone

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

@@ -50,11 +50,13 @@ class Parser(hyperparams.Config):
   small_instance_weight: float = 3.0
   dtype = 'float32'
 
+
 @dataclasses.dataclass
 class DataDecoder(common.DataDecoder):
   """Data decoder config."""
   simple_decoder: common.TfExampleDecoder = common.TfExampleDecoder()
 
+
 @dataclasses.dataclass
 class DataConfig(cfg.DataConfig):
   """Input config for training."""
@@ -62,6 +64,7 @@ class DataConfig(cfg.DataConfig):
   parser: Parser = Parser()
   file_type: str = 'tfrecord'
 
+
 @dataclasses.dataclass
 class PanopticDeeplabHead(hyperparams.Config):
   """Panoptic Deeplab head config."""
@@ -75,16 +78,19 @@ class PanopticDeeplabHead(hyperparams.Config):
   low_level_num_filters: Union[List[int], Tuple[int]] = (64, 32)
   fusion_num_output_filters: int = 256
 
+
 @dataclasses.dataclass
 class SemanticHead(PanopticDeeplabHead):
   """Semantic head config."""
   prediction_kernel_size: int = 1
 
+
 @dataclasses.dataclass
 class InstanceHead(PanopticDeeplabHead):
   """Instance head config."""
   prediction_kernel_size: int = 1
 
+
 @dataclasses.dataclass
 class PanopticDeeplabPostProcessor(hyperparams.Config):
   """Panoptic Deeplab PostProcessing config."""
@@ -99,6 +105,7 @@ class PanopticDeeplabPostProcessor(hyperparams.Config):
   keep_k_centers: int = 200
   rescale_predictions: bool = True
 
+
 @dataclasses.dataclass
 class PanopticDeeplab(hyperparams.Config):
   """Panoptic Deeplab model config."""
@@ -116,6 +123,7 @@ class PanopticDeeplab(hyperparams.Config):
   generate_panoptic_masks: bool = True
   post_processor: PanopticDeeplabPostProcessor = PanopticDeeplabPostProcessor()
 
+
 @dataclasses.dataclass
 class Losses(hyperparams.Config):
   label_smoothing: float = 0.0
@@ -127,6 +135,7 @@ class Losses(hyperparams.Config):
   center_heatmap_loss_weight: float = 200
   center_offset_loss_weight: float = 0.01
 
+
 @dataclasses.dataclass
 class Evaluation(hyperparams.Config):
   """ Evaluation config """
@@ -141,6 +150,7 @@ class Evaluation(hyperparams.Config):
   report_per_class_iou: bool = False
   report_train_mean_iou: bool = True  # Turning this off can speed up training.
 
+
 @dataclasses.dataclass
 class PanopticDeeplabTask(cfg.TaskConfig):
   model: PanopticDeeplab = PanopticDeeplab()
@@ -175,10 +185,9 @@ def panoptic_deeplab_coco() -> cfg.ExperimentConfig:
   output_stride = 16
   aspp_dilation_rates = [6, 12, 18]
   multigrid = [1, 2, 4]
-  stem_type = 'v0'
+  stem_type = 'v1'
   level = int(np.math.log2(output_stride))
 
-
   config = cfg.ExperimentConfig(
       runtime=cfg.RuntimeConfig(
           mixed_precision_dtype='bfloat16', enable_xla=True),
@@ -191,9 +200,12 @@ def panoptic_deeplab_coco() -> cfg.ExperimentConfig:
               backbone=backbones.Backbone(
                   type='dilated_resnet', dilated_resnet=backbones.DilatedResNet(
                       model_id=50,
+                      stem_type=stem_type,
                       output_stride=output_stride,
                       multigrid=multigrid,
-                      stem_type=stem_type)),
+                      se_ratio=0.25,
+                      last_stage_repeats=1,
+                      stochastic_depth_drop_rate=0.2)),
               decoder=decoders.Decoder(
                   type='aspp',
                   aspp=decoders.ASPP(
@@ -201,6 +213,7 @@ def panoptic_deeplab_coco() -> cfg.ExperimentConfig:
                       num_filters=256,
                       pool_kernel_size=input_size[:2],
                       dilation_rates=aspp_dilation_rates,
+                      use_depthwise_convolution=True,
                       dropout_rate=0.1)),
               semantic_head=SemanticHead(
                   level=level,