Make both input types work

official/projects/detr/configs/detr.py

@@ -58,6 +58,7 @@ class DataConfig(cfg.DataConfig):
 
 @dataclasses.dataclass
 class Losses(hyperparams.Config):
+  class_offset: int = 0
   lambda_cls: float = 1.0
   lambda_box: float = 5.0
   lambda_giou: float = 2.0
@@ -101,7 +102,7 @@ def detr_coco() -> cfg.ExperimentConfig:
   decay_at = train_steps - 100 * num_steps_per_epoch  # 400 epochs
   config = cfg.ExperimentConfig(
       task=DetrTask(
-          init_checkpoint='gs://ghpark-imagenet-tfrecord/ckpt/resnet50_imagenet',
+          init_checkpoint='gs://tf_model_garden/vision/resnet50_imagenet/ckpt-62400',
           init_checkpoint_modules='backbone',
           model=Detr(
               num_classes=81,
@@ -109,7 +110,6 @@ def detr_coco() -> cfg.ExperimentConfig:
               norm_activation=common.NormActivation(use_sync_bn=False)),
           losses=Losses(),
           train_data=coco.COCODataConfig(
-              file_type='tfrecord',
               tfds_name='coco/2017',
               tfds_split='train',
               is_training=True,
@@ -117,7 +117,6 @@ def detr_coco() -> cfg.ExperimentConfig:
               shuffle_buffer_size=1000,
           ),
           validation_data=coco.COCODataConfig(
-              file_type='tfrecord',
               tfds_name='coco/2017',
               tfds_split='validation',
               is_training=False,
@@ -159,7 +158,7 @@ def detr_coco() -> cfg.ExperimentConfig:
       ])
   return config
 
-COCO_INPUT_PATH_BASE = 'gs://ghpark-tfrecords/coco'
+COCO_INPUT_PATH_BASE = ''
 COCO_TRAIN_EXAMPLES = 118287
 COCO_VAL_EXAMPLES = 5000
 
@@ -173,7 +172,7 @@ def detr_coco() -> cfg.ExperimentConfig:
   decay_at = train_steps - 100 * steps_per_epoch  # 200 epochs
   config = cfg.ExperimentConfig(
       task=DetrTask(
-          init_checkpoint='gs://ghpark-imagenet-tfrecord/ckpt/resnet50_imagenet',
+          init_checkpoint='gs://tf_model_garden/vision/resnet50_imagenet/ckpt-62400',
           init_checkpoint_modules='backbone',
           annotation_file=os.path.join(COCO_INPUT_PATH_BASE,
                                        'instances_val2017.json'),
@@ -227,3 +226,71 @@ def detr_coco() -> cfg.ExperimentConfig:
           'task.train_data.is_training != None',
       ])
   return config
+
+@exp_factory.register_config_factory('detr_coco_tfds')
+def detr_coco() -> cfg.ExperimentConfig:
+  """Config to get results that matches the paper."""
+  train_batch_size = 64
+  eval_batch_size = 64
+  steps_per_epoch = COCO_TRAIN_EXAMPLES // train_batch_size
+  train_steps = 300 * steps_per_epoch  # 300 epochs
+  decay_at = train_steps - 100 * steps_per_epoch  # 200 epochs
+  config = cfg.ExperimentConfig(
+      task=DetrTask(
+          init_checkpoint='gs://tf_model_garden/vision/resnet50_imagenet/ckpt-62400',
+          init_checkpoint_modules='backbone',
+          model=Detr(
+              num_classes=81,
+              input_size=[1333, 1333, 3],
+              norm_activation=common.NormActivation(use_sync_bn=False)),
+          losses=Losses(
+              class_offset=1
+          ),
+          train_data=DataConfig(
+              tfds_name='coco/2017',
+              tfds_split='train',
+              is_training=True,
+              global_batch_size=train_batch_size,
+              shuffle_buffer_size=1000,
+          ),
+          validation_data=DataConfig(
+              tfds_name='coco/2017',
+              tfds_split='validation',
+              is_training=False,
+              global_batch_size=eval_batch_size,
+              drop_remainder=False
+          )
+      ),
+      trainer=cfg.TrainerConfig(
+          train_steps=train_steps,
+          validation_steps=COCO_VAL_EXAMPLES // eval_batch_size,
+          steps_per_loop=steps_per_epoch,
+          summary_interval=steps_per_epoch,
+          checkpoint_interval=steps_per_epoch,
+          validation_interval=5*steps_per_epoch,
+          max_to_keep=1,
+          best_checkpoint_export_subdir='best_ckpt',
+          best_checkpoint_eval_metric='AP',
+          optimizer_config=optimization.OptimizationConfig({
+              'optimizer': {
+                  'type': 'detr_adamw',
+                  'detr_adamw': {
+                      'weight_decay_rate': 1e-4,
+                      'global_clipnorm': 0.1,
+                      # Avoid AdamW legacy behavior.
+                      'gradient_clip_norm': 0.0
+                  }
+              },
+              'learning_rate': {
+                  'type': 'stepwise',
+                  'stepwise': {
+                      'boundaries': [decay_at],
+                      'values': [0.0001, 1.0e-05]
+                  }
+              },
+              })
+          ),
+      restrictions=[
+          'task.train_data.is_training != None',
+      ])
+  return config
\ No newline at end of file

official/projects/detr/dataloaders/coco.py

@@ -27,7 +27,6 @@ from official.vision.ops import preprocess_ops
 @dataclasses.dataclass
 class COCODataConfig(cfg.DataConfig):
   """Data config for COCO."""
-  file_type: str = 'tfrecord'
   output_size: Tuple[int, int] = (1333, 1333)
   max_num_boxes: int = 100
   resize_scales: Tuple[int, ...] = (

official/projects/detr/dataloaders/detr_input.py

@@ -31,34 +31,28 @@ class Parser(parser.Parser):
   """Parse an image and its annotations into a dictionary of tensors."""
 
   def __init__(self,
+               class_offset: int = 0,
                output_size: Tuple[int, int] = (1333, 1333),
                max_num_boxes: int = 100,
                resize_scales: Tuple[int, ...] = RESIZE_SCALES,
                aug_rand_hflip=True):
+    self._class_offset = class_offset
     self._output_size = output_size
     self._max_num_boxes = max_num_boxes
     self._resize_scales = resize_scales
     self._aug_rand_hflip = aug_rand_hflip
-    
               
   def _parse_train_data(self, data):
     """Parses data for training and evaluation."""
-    #classes = data['groundtruth_classes'] + 1
-    classes = data['groundtruth_classes']
+    classes = data['groundtruth_classes'] + self._class_offset
     boxes = data['groundtruth_boxes']
     is_crowd = data['groundtruth_is_crowd']
 
     # Gets original image.
     image = data['image']
 
-    # Apply autoaug or randaug.
-    #if self._augmenter is not None:
-    #  image, boxes = self._augmenter.distort_with_boxes(image, boxes)
-
     # Normalizes image with mean and std pixel values.
     image = preprocess_ops.normalize_image(image)
-
-    
     image, boxes, _ = preprocess_ops.random_horizontal_flip(image, boxes)
 
     do_crop = tf.greater(tf.random.uniform([]), 0.5)

official/projects/detr/tasks/detection.py

@@ -31,6 +31,7 @@ from official.vision.dataloaders import tf_example_decoder
 from official.vision.dataloaders import tfds_factory
 from official.vision.dataloaders import tf_example_label_map_decoder
 from official.projects.detr.dataloaders import detr_input
+from official.projects.detr.dataloaders import coco
 from official.vision.modeling import backbones
 
 @task_factory.register_task_cls(detr_cfg.DetrTask)
@@ -84,41 +85,39 @@ class DectectionTask(base_task.Task):
     logging.info('Finished loading pretrained checkpoint from %s',
                  ckpt_dir_or_file)
 
-  """def build_inputs(self,
-                   params: detr_cfg.DataConfig,
-                   input_context: Optional[tf.distribute.InputContext] = None):
-    return coco.COCODataLoader(params).load(input_context)"""
-  
   def build_inputs(self,
                    params,
                    input_context: Optional[tf.distribute.InputContext] = None):
     """Build input dataset."""
-
-    if params.tfds_name:
-      decoder = tfds_factory.get_detection_decoder(params.tfds_name)
+    if type(params) is coco.COCODataConfig:
+      dataset = coco.COCODataLoader(params).load(input_context)
     else:
-      decoder_cfg = params.decoder.get()
-      if params.decoder.type == 'simple_decoder':
-        decoder = tf_example_decoder.TfExampleDecoder(
-            regenerate_source_id=decoder_cfg.regenerate_source_id)
-      elif params.decoder.type == 'label_map_decoder':
-        decoder = tf_example_label_map_decoder.TfExampleDecoderLabelMap(
-            label_map=decoder_cfg.label_map,
-            regenerate_source_id=decoder_cfg.regenerate_source_id)
+      if params.tfds_name:
+        decoder = tfds_factory.get_detection_decoder(params.tfds_name)
       else:
-        raise ValueError('Unknown decoder type: {}!'.format(
-            params.decoder.type))
-    
-    parser = detr_input.Parser(
-        output_size=self._task_config.model.input_size[:2],
-    )
-
-    reader = input_reader_factory.input_reader_generator(
-        params,
-        dataset_fn=dataset_fn.pick_dataset_fn(params.file_type),
-        decoder_fn=decoder.decode,
-        parser_fn=parser.parse_fn(params.is_training))
-    dataset = reader.read(input_context=input_context)
+        decoder_cfg = params.decoder.get()
+        if params.decoder.type == 'simple_decoder':
+          decoder = tf_example_decoder.TfExampleDecoder(
+              regenerate_source_id=decoder_cfg.regenerate_source_id)
+        elif params.decoder.type == 'label_map_decoder':
+          decoder = tf_example_label_map_decoder.TfExampleDecoderLabelMap(
+              label_map=decoder_cfg.label_map,
+              regenerate_source_id=decoder_cfg.regenerate_source_id)
+        else:
+          raise ValueError('Unknown decoder type: {}!'.format(
+              params.decoder.type))
+      
+      parser = detr_input.Parser(
+          class_offset=self._task_config.losses.class_offset,
+          output_size=self._task_config.model.input_size[:2],
+      )
+
+      reader = input_reader_factory.input_reader_generator(
+          params,
+          dataset_fn=dataset_fn.pick_dataset_fn(params.file_type),
+          decoder_fn=decoder.decode,
+          parser_fn=parser.parse_fn(params.is_training))
+      dataset = reader.read(input_context=input_context)
 
     return dataset