Mask R-CNN model added to models/research/mlperf_object_detection/Mask_RCNN (#4678)

* Create README.md

* readme changed

* readme changed

* ResNet backbone completed.

* FPN added

* Create README.md

* initial commit

* files removed

* initial commit

* protobuf file removed

Author: MehdiShz

research/mlperf_object_detection/Mask_RCNN/README.md

@@ -0,0 +1 @@
+Mask RCNN Implimentation adopted from models/research/object_detection/

research/mlperf_object_detection/Mask_RCNN/configs/e2e_mask_rcnn_R-50-C4_atrous.config

@@ -0,0 +1,170 @@
+# Mask R-CNN with Resnet-50 (v1), Atrous version
+# Configured for MSCOCO Dataset.
+# Users should configure the fine_tune_checkpoint field in the train config as
+# well as the label_map_path and input_path fields in the train_input_reader and
+# eval_input_reader. Search for "PATH_TO_BE_CONFIGURED" to find the fields that
+# should be configured.
+
+model {
+  faster_rcnn {
+    num_classes: 81
+    image_resizer {
+      keep_aspect_ratio_resizer {
+        min_dimension: 800
+        max_dimension: 1365
+      }
+    }
+    number_of_stages: 3
+    feature_extractor {
+      type: 'faster_rcnn_resnet50'
+      first_stage_features_stride: 8
+    }
+    first_stage_anchor_generator {
+      grid_anchor_generator {
+        scales: [0.125, 0.25, 0.5, 1.0, 2.0]	# base size=256**2 => anchor sizes=32 64 128 256 512
+        aspect_ratios: [0.5, 1.0, 2.0]
+        height_stride: 8
+        width_stride: 8
+      }
+    }
+    first_stage_atrous_rate: 2
+    first_stage_box_predictor_conv_hyperparams {
+      op: CONV
+      regularizer {
+        l2_regularizer {
+          weight: 0.0
+        }
+      }
+      initializer {
+        truncated_normal_initializer {
+          stddev: 0.01
+        }
+      }
+    }
+    first_stage_nms_score_threshold: 0.0
+    first_stage_nms_iou_threshold: 0.7
+    first_stage_max_proposals: 512
+    first_stage_localization_loss_weight: 2.0
+    first_stage_objectness_loss_weight: 1.0
+    initial_crop_size: 14
+    maxpool_kernel_size: 2
+    maxpool_stride: 2
+    second_stage_batch_size: 512
+    second_stage_box_predictor {
+      mask_rcnn_box_predictor {
+        use_dropout: false
+        dropout_keep_probability: 1.0
+        predict_instance_masks: true
+        mask_height: 14
+        mask_width: 14
+        mask_prediction_conv_depth: 0
+        mask_prediction_num_conv_layers: 3	#from mask rcnn heads 
+        fc_hyperparams {
+          op: FC
+          regularizer {
+            l2_regularizer {
+              weight: 0.0
+            }
+          }
+          initializer {
+            variance_scaling_initializer {
+              factor: 1.0
+              uniform: true
+              mode: FAN_AVG
+            }
+          }
+        }
+        conv_hyperparams {
+          op: CONV
+          regularizer {
+            l2_regularizer {
+              weight: 0.0
+            }
+          }
+          initializer {
+            truncated_normal_initializer {
+              stddev: 0.01
+            }
+          }
+        }
+      }
+    }
+    second_stage_post_processing {
+      batch_non_max_suppression {
+        score_threshold: 0.0
+        iou_threshold: 0.6
+        max_detections_per_class: 2000
+        max_total_detections: 2000
+      }
+      score_converter: SOFTMAX
+    }
+    second_stage_localization_loss_weight: 2.0
+    second_stage_classification_loss_weight: 1.0
+    second_stage_mask_prediction_loss_weight: 4.0
+  }
+}
+
+train_config: {
+  batch_size: 4
+  optimizer {
+    momentum_optimizer: {
+      learning_rate: {
+        manual_step_learning_rate {
+          initial_learning_rate: 0.01
+          schedule {
+            step: 120000
+            learning_rate: .001
+          }
+          schedule {
+            step: 160000
+            learning_rate: .0001
+          }
+        }
+      }
+      momentum_optimizer_value: 0.9
+    }
+    use_moving_average: false
+  }
+  gradient_clipping_by_norm: 10.0
+  #fine_tune_checkpoint: "/home/mehdisharif/data/coco/resnet_v1_50.ckpt"
+  #from_detection_checkpoint: True
+  # Note: The below line limits the training process to 200K steps, which we
+  # empirically found to be sufficient enough to train the pets dataset. This
+  # effectively bypasses the learning rate schedule (the learning rate will
+  # never decay). Remove the below line to train indefinitely.
+  num_steps: 20000000
+  data_augmentation_options {
+    random_horizontal_flip {
+    }
+  }
+}
+
+train_input_reader: {
+  tf_record_input_reader {
+    input_path: "/home/mehdisharif/data/coco/output2017/coco_train.record"
+  }
+  label_map_path: "/home/mehdisharif/data/coco/output2017/mscoco_label_map.pbtxt"
+  load_instance_masks: true
+  mask_type: PNG_MASKS
+}
+
+eval_config: {
+  metrics_set: ['coco_detection_metrics', 'coco_mask_metrics']
+  num_examples: 50
+  # Note: The below line limits the evaluation process to 10 evaluations.
+  # Remove the below line to evaluate indefinitely.
+  max_evals: 1
+  num_visualizations: 50
+  eval_interval_secs: 120
+}
+
+eval_input_reader: {
+  tf_record_input_reader {
+    input_path: "/home/mehdisharif/data/coco/output2017/coco_val.record"
+  }
+  label_map_path: "/home/mehdisharif/data/coco/output2017/mscoco_label_map.pbtxt"
+  load_instance_masks: true
+  mask_type: PNG_MASKS
+  shuffle: false
+  num_readers: 1
+}

research/mlperf_object_detection/Mask_RCNN/configs/mask_rcnn_resnet50_atrous_coco.config

@@ -0,0 +1,169 @@
+# Mask R-CNN with Resnet-50 (v1), Atrous version
+# Configured for MSCOCO Dataset.
+# Users should configure the fine_tune_checkpoint field in the train config as
+# well as the label_map_path and input_path fields in the train_input_reader and
+# eval_input_reader. Search for "PATH_TO_BE_CONFIGURED" to find the fields that
+# should be configured.
+
+model {
+  faster_rcnn {
+    num_classes: 90
+    image_resizer {
+      keep_aspect_ratio_resizer {
+        min_dimension: 800
+        max_dimension: 1365
+      }
+    }
+    number_of_stages: 3
+    feature_extractor {
+      type: 'faster_rcnn_resnet50'
+      first_stage_features_stride: 8
+    }
+    first_stage_anchor_generator {
+      grid_anchor_generator {
+        scales: [0.25, 0.5, 1.0, 2.0]
+        aspect_ratios: [0.5, 1.0, 2.0]
+        height_stride: 8
+        width_stride: 8
+      }
+    }
+    first_stage_atrous_rate: 2
+    first_stage_box_predictor_conv_hyperparams {
+      op: CONV
+      regularizer {
+        l2_regularizer {
+          weight: 0.0
+        }
+      }
+      initializer {
+        truncated_normal_initializer {
+          stddev: 0.01
+        }
+      }
+    }
+    first_stage_nms_score_threshold: 0.0
+    first_stage_nms_iou_threshold: 0.7
+    first_stage_max_proposals: 300
+    first_stage_localization_loss_weight: 2.0
+    first_stage_objectness_loss_weight: 1.0
+    initial_crop_size: 14
+    maxpool_kernel_size: 2
+    maxpool_stride: 2
+    second_stage_box_predictor {
+      mask_rcnn_box_predictor {
+        use_dropout: false
+        dropout_keep_probability: 1.0
+        predict_instance_masks: true
+        mask_height: 33
+        mask_width: 33
+        mask_prediction_conv_depth: 0
+        mask_prediction_num_conv_layers: 4
+        fc_hyperparams {
+          op: FC
+          regularizer {
+            l2_regularizer {
+              weight: 0.0
+            }
+          }
+          initializer {
+            variance_scaling_initializer {
+              factor: 1.0
+              uniform: true
+              mode: FAN_AVG
+            }
+          }
+        }
+        conv_hyperparams {
+          op: CONV
+          regularizer {
+            l2_regularizer {
+              weight: 0.0
+            }
+          }
+          initializer {
+            truncated_normal_initializer {
+              stddev: 0.01
+            }
+          }
+        }
+      }
+    }
+    second_stage_post_processing {
+      batch_non_max_suppression {
+        score_threshold: 0.0
+        iou_threshold: 0.6
+        max_detections_per_class: 100
+        max_total_detections: 300
+      }
+      score_converter: SOFTMAX
+    }
+    second_stage_localization_loss_weight: 2.0
+    second_stage_classification_loss_weight: 1.0
+    second_stage_mask_prediction_loss_weight: 4.0
+  }
+}
+
+train_config: {
+  batch_size: 2
+  optimizer {
+    momentum_optimizer: {
+      learning_rate: {
+        manual_step_learning_rate {
+          initial_learning_rate: 0.0003
+          schedule {
+            step: 900000
+            learning_rate: .00003
+          }
+          schedule {
+            step: 1200000
+            learning_rate: .000003
+          }
+        }
+      }
+      momentum_optimizer_value: 0.9
+    }
+    use_moving_average: false
+  }
+  gradient_clipping_by_norm: 10.0
+  #fine_tune_checkpoint: ""
+  from_detection_checkpoint: false
+  # Note: The below line limits the training process to 200K steps, which we
+  # empirically found to be sufficient enough to train the pets dataset. This
+  # effectively bypasses the learning rate schedule (the learning rate will
+  # never decay). Remove the below line to train indefinitely.
+  #num_steps: 200000
+  data_augmentation_options {
+    random_horizontal_flip {
+    }
+  }
+}
+
+train_input_reader: {
+  tf_record_input_reader {
+    input_path: "PATH_TO_BE_CONFIGURED/coco_train.record"
+  }
+  label_map_path: "PATH_TO_BE_CONFIGURED/mscoco_label_map.pbtxt"
+  load_instance_masks: true
+  mask_type: PNG_MASKS
+}
+
+eval_config: {
+  metrics_set: ['coco_detection_metrics', 'coco_mask_metrics']
+  num_examples: 50
+  # Note: The below line limits the evaluation process to 10 evaluations.
+  # Remove the below line to evaluate indefinitely.
+  max_evals: 1
+  num_visualizations: 50
+  eval_interval_secs: 120
+}
+
+eval_input_reader: {
+  tf_record_input_reader {
+    input_path: "PATH_TO_BE_CONFIGURED/coco_val.record"
+  }
+  label_map_path: "PATH_TO_BE_CONFIGURED/mscoco_label_map.pbtxt"
+  load_instance_masks: true
+  mask_type: PNG_MASKS
+  shuffle: true
+  num_readers: 1
+}