Upload 2 files

maskdino_SwinL_bs16_50ep_4s_dowsample1_2048_x-ray_11diseases.yaml

@@ -0,0 +1,84 @@
+_BASE_: ../Base-COCO-InstanceSegmentation.yaml
+DATASETS:
+  TRAIN: ("x-ray-11classes",)
+  TEST: ("x-ray-11classes",)
+MODEL:
+  META_ARCHITECTURE: "MaskDINO"
+  BACKBONE:
+    NAME: "D2SwinTransformer"
+  SWIN:
+    EMBED_DIM: 192
+    DEPTHS: [ 2, 2, 18, 2 ]
+    NUM_HEADS: [ 6, 12, 24, 48 ]
+    WINDOW_SIZE: 12
+    APE: False
+    DROP_PATH_RATE: 0.3
+    PATCH_NORM: True
+    PRETRAIN_IMG_SIZE: 384
+  WEIGHTS: "swin_large_patch4_window12_384_22k.pkl"
+  PIXEL_MEAN: [ 123.675, 116.280, 103.530 ]
+  PIXEL_STD: [ 58.395, 57.120, 57.375 ]
+  # head
+  SEM_SEG_HEAD:
+    NAME: "MaskDINOHead"
+    IGNORE_VALUE: 255
+    NUM_CLASSES: 11
+    LOSS_WEIGHT: 1.0
+    CONVS_DIM: 256
+    MASK_DIM: 256
+    NORM: "GN"
+    # pixel decoder
+    PIXEL_DECODER_NAME: "MaskDINOEncoder"
+    DIM_FEEDFORWARD: 2048
+    NUM_FEATURE_LEVELS: 4
+    TOTAL_NUM_FEATURE_LEVELS: 5
+    IN_FEATURES: ["res2", "res3", "res4", "res5"]
+    DEFORMABLE_TRANSFORMER_ENCODER_IN_FEATURES: ["res2","res3", "res4", "res5"]
+    COMMON_STRIDE: 4
+    TRANSFORMER_ENC_LAYERS: 6
+    FEATURE_ORDER: "low2high"
+  MaskDINO:
+    TRANSFORMER_DECODER_NAME: "MaskDINODecoder"
+    DEEP_SUPERVISION: True
+    NO_OBJECT_WEIGHT: 0.1
+    CLASS_WEIGHT: 4.0
+    MASK_WEIGHT: 5.0
+    DICE_WEIGHT: 5.0
+    BOX_WEIGHT: 5.0
+    GIOU_WEIGHT: 2.0
+    HIDDEN_DIM: 256
+    NUM_OBJECT_QUERIES: 300
+    NHEADS: 8
+    DROPOUT: 0.0
+    DIM_FEEDFORWARD: 2048
+    ENC_LAYERS: 0
+    PRE_NORM: False
+    ENFORCE_INPUT_PROJ: False
+    SIZE_DIVISIBILITY: 32
+    DEC_LAYERS: 9  # 9+1, 9 decoder layers, add one for the loss on learnable query
+    TRAIN_NUM_POINTS: 12544
+    OVERSAMPLE_RATIO: 3.0
+    IMPORTANCE_SAMPLE_RATIO: 0.75
+    EVAL_FLAG: 1
+    INITIAL_PRED: True
+    TWO_STAGE: True
+    DN: "seg"
+    DN_NUM: 100
+    INITIALIZE_BOX_TYPE: 'bitmask'
+    TEST:
+      SEMANTIC_ON: False
+      INSTANCE_ON: True
+      PANOPTIC_ON: False
+      OVERLAP_THRESHOLD: 0.8
+      OBJECT_MASK_THRESHOLD: 0.25
+
+SOLVER:
+  IMS_PER_BATCH: 2
+  BASE_LR: 0.0001
+  STEPS: (36000, 39000)
+  MAX_ITER: 40000
+  AMP:
+    ENABLED: True
+TEST:
+  EVAL_PERIOD: 39999
+#  EVAL_FLAG: 1

maskdino_SwinL_bs16_50ep_4s_dowsample1_2048_x-ray_bone_loss_1diseases.yaml

@@ -0,0 +1,84 @@
+_BASE_: ../Base-COCO-InstanceSegmentation.yaml
+DATASETS:
+  TRAIN: ("bone_loss_1classes_train",)
+  TEST: ("bone_loss_1classes_val",)
+MODEL:
+  META_ARCHITECTURE: "MaskDINO"
+  BACKBONE:
+    NAME: "D2SwinTransformer"
+  SWIN:
+    EMBED_DIM: 192
+    DEPTHS: [ 2, 2, 18, 2 ]
+    NUM_HEADS: [ 6, 12, 24, 48 ]
+    WINDOW_SIZE: 12
+    APE: False
+    DROP_PATH_RATE: 0.3
+    PATCH_NORM: True
+    PRETRAIN_IMG_SIZE: 384
+  WEIGHTS: "swin_large_patch4_window12_384_22k.pkl"
+  PIXEL_MEAN: [ 123.675, 116.280, 103.530 ]
+  PIXEL_STD: [ 58.395, 57.120, 57.375 ]
+  # head
+  SEM_SEG_HEAD:
+    NAME: "MaskDINOHead"
+    IGNORE_VALUE: 255
+    NUM_CLASSES: 1
+    LOSS_WEIGHT: 1.0
+    CONVS_DIM: 256
+    MASK_DIM: 256
+    NORM: "GN"
+    # pixel decoder
+    PIXEL_DECODER_NAME: "MaskDINOEncoder"
+    DIM_FEEDFORWARD: 2048
+    NUM_FEATURE_LEVELS: 4
+    TOTAL_NUM_FEATURE_LEVELS: 5
+    IN_FEATURES: ["res2", "res3", "res4", "res5"]
+    DEFORMABLE_TRANSFORMER_ENCODER_IN_FEATURES: ["res2","res3", "res4", "res5"]
+    COMMON_STRIDE: 4
+    TRANSFORMER_ENC_LAYERS: 6
+    FEATURE_ORDER: "low2high"
+  MaskDINO:
+    TRANSFORMER_DECODER_NAME: "MaskDINODecoder"
+    DEEP_SUPERVISION: True
+    NO_OBJECT_WEIGHT: 0.1
+    CLASS_WEIGHT: 4.0
+    MASK_WEIGHT: 5.0
+    DICE_WEIGHT: 5.0
+    BOX_WEIGHT: 5.0
+    GIOU_WEIGHT: 2.0
+    HIDDEN_DIM: 256
+    NUM_OBJECT_QUERIES: 300
+    NHEADS: 8
+    DROPOUT: 0.0
+    DIM_FEEDFORWARD: 2048
+    ENC_LAYERS: 0
+    PRE_NORM: False
+    ENFORCE_INPUT_PROJ: False
+    SIZE_DIVISIBILITY: 32
+    DEC_LAYERS: 9  # 9+1, 9 decoder layers, add one for the loss on learnable query
+    TRAIN_NUM_POINTS: 12544
+    OVERSAMPLE_RATIO: 3.0
+    IMPORTANCE_SAMPLE_RATIO: 0.75
+    EVAL_FLAG: 1
+    INITIAL_PRED: True
+    TWO_STAGE: True
+    DN: "seg"
+    DN_NUM: 100
+    INITIALIZE_BOX_TYPE: 'bitmask'
+    TEST:
+      SEMANTIC_ON: False
+      INSTANCE_ON: True
+      PANOPTIC_ON: False
+      OVERLAP_THRESHOLD: 0.8
+      OBJECT_MASK_THRESHOLD: 0.25
+
+SOLVER:
+  IMS_PER_BATCH: 2
+  BASE_LR: 0.0001
+  STEPS: (36000, 39000)
+  MAX_ITER: 40000
+  AMP:
+    ENABLED: True
+TEST:
+  EVAL_PERIOD: 5000
+#  EVAL_FLAG: 1