mitosis detection（三）训练成功

于 2020-03-03 18:16:32 发布
阅读量267 收藏
点赞数
分类专栏： mitosis caffe 项目设计
版权声明：本文为博主原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。
本文链接：https://blog.csdn.net/qq_41574648/article/details/104638131
版权
mitosis 同时被 3 个专栏收录
8 篇文章 0 订阅
订阅专栏
caffe
8 篇文章 0 订阅
订阅专栏
项目设计
3 篇文章 0 订阅
订阅专栏 
./experiments/scripts/faster_rcnn_end2end_12.sh 6 VGG_CNN_M_1024_Scale mitos --set EXP_DIR XX
+ set -e
+ export PYTHONUNBUFFERED=True
+ PYTHONUNBUFFERED=True
+ GPU_ID=6
+ NET=VGG_CNN_M_1024_Scale
+ NET_lc=vgg_cnn_m_1024_scale
+ DATASET=mitos
+ array=($@)
+ len=6
+ EXTRA_ARGS='--set EXP_DIR XX'
+ EXTRA_ARGS_SLUG=--set_EXP_DIR_XX
+ case $DATASET in
+ TRAIN_IMDB=mitos_2012_train
+ TEST_IMDB=mitos_2012_test
+ PT_DIR=mitos
+ ITERS=150000
++ date +%Y-%m-%d_%H-%M-%S
+ LOG=experiments/logs/faster_rcnn_end2end_VGG_CNN_M_1024_Scale_--set_EXP_DIR_XX.txt.2020-03-01_13-13-36
+ exec
++ tee -a experiments/logs/faster_rcnn_end2end_VGG_CNN_M_1024_Scale_--set_EXP_DIR_XX.txt.2020-03-01_13-13-3                             6
+ echo Logging output to experiments/logs/faster_rcnn_end2end_VGG_CNN_M_1024_Scale_--set_EXP_DIR_XX.txt.202                             0-03-01_13-13-36
Logging output to experiments/logs/faster_rcnn_end2end_VGG_CNN_M_1024_Scale_--set_EXP_DIR_XX.txt.2020-03-01                             _13-13-36
+ ./tools/train_net.py --gpu 6 --solver models/mitos/VGG_CNN_M_1024_Scale/faster_rcnn_end2end/solver.protot                             xt --weights data/imagenet_models/VGG_CNN_M_1024_Scale.v2.caffemodel --imdb mitos_2012_train --iters 150000                              --cfg experiments/cfgs/faster_rcnn_end2end_2012.yml --set EXP_DIR XX
Called with args:
Namespace(cfg_file='experiments/cfgs/faster_rcnn_end2end_2012.yml', gpu_id=6, imdb_name='mitos_2012_train',                              max_iters=150000, pretrained_model='data/imagenet_models/VGG_CNN_M_1024_Scale.v2.caffemodel', randomize=Fa                             lse, set_cfgs=['EXP_DIR', 'XX'], solver='models/mitos/VGG_CNN_M_1024_Scale/faster_rcnn_end2end/solver.proto                             txt')
Using config:
{'DATA_DIR': '/home/lyh/mitosis_DET/DeepMitosis/DeepDetection/data',
 'DEDUP_BOXES': 0.0625,
 'EPS': 1e-14,
 'EXP_DIR': 'XX',
 'GPU_ID': 6,
 'MATLAB': 'matlab',
 'MODELS_DIR': '/home/lyh/mitosis_DET/DeepMitosis/DeepDetection/models/pascal_voc',
 'PIXEL_MEANS': array([[[102.9801, 115.9465, 122.7717]]]),
 'RNG_SEED': 3,
 'ROOT_DIR': '/home/lyh/mitosis_DET/DeepMitosis/DeepDetection',
 'TEST': {'BBOX_REG': True,
          'HAS_RPN': True,
          'MAX_SIZE': 4168,
          'NMS': 0.3,
          'PROPOSAL_METHOD': 'selective_search',
          'RPN_MIN_SIZE': 16,
          'RPN_NMS_THRESH': 0.7,
          'RPN_POST_NMS_TOP_N': 300,
          'RPN_PRE_NMS_TOP_N': 6000,
          'SCALES': [4168],
          'SVM': False},
 'TRAIN': {'ASPECT_GROUPING': True,
           'BATCH_SIZE': 128,
           'BBOX_INSIDE_WEIGHTS': [1.0, 1.0, 1.0, 1.0],
           'BBOX_NORMALIZE_MEANS': [0.0, 0.0, 0.0, 0.0],
           'BBOX_NORMALIZE_STDS': [0.1, 0.1, 0.2, 0.2],
           'BBOX_NORMALIZE_TARGETS': True,
           'BBOX_NORMALIZE_TARGETS_PRECOMPUTED': True,
           'BBOX_REG': True,
           'BBOX_THRESH': 0.5,
           'BG_THRESH_HI': 0.5,
           'BG_THRESH_LO': 0.0,
           'FG_FRACTION': 0.25,
           'FG_THRESH': 0.5,
           'HAS_RPN': True,
           'IMS_PER_BATCH': 1,
           'MAX_SIZE': 1024,
           'PROPOSAL_METHOD': 'gt',
           'RPN_BATCHSIZE': 256,
           'RPN_BBOX_INSIDE_WEIGHTS': [1.0, 1.0, 1.0, 1.0],
           'RPN_CLOBBER_POSITIVES': False,
           'RPN_FG_FRACTION': 0.5,
           'RPN_MIN_SIZE': 16,
           'RPN_NEGATIVE_OVERLAP': 0.3,
           'RPN_NMS_THRESH': 0.7,
           'RPN_POSITIVE_OVERLAP': 0.7,
           'RPN_POSITIVE_WEIGHT': -1.0,
           'RPN_POST_NMS_TOP_N': 2000,
           'RPN_PRE_NMS_TOP_N': 12000,
           'SCALES': [1024],
           'SNAPSHOT_INFIX': '',
           'SNAPSHOT_ITERS': 10000,
           'USE_FLIPPED': True,
           'USE_PREFETCH': False},
 'USE_GPU_NMS': True}
Loaded dataset `mitos_2012_train` for training
Set proposal method: gt
Appending horizontally-flipped training examples...
wrote gt roidb to /home/lyh/mitosis_DET/DeepMitosis/DeepDetection/data/cache/mitos_2012_train_gt_roidb.pkl
done
Preparing training data...
done
60076 roidb entries
Output will be saved to `/home/lyh/mitosis_DET/DeepMitosis/DeepDetection/output/XX/mitos_2012_train`
Filtered 0 roidb entries: 60076 -> 60076
Computing bounding-box regression targets...
bbox target means:
[[0. 0. 0. 0.]
 [0. 0. 0. 0.]]
[0. 0. 0. 0.]
bbox target stdevs:
[[0.1 0.1 0.2 0.2]
 [0.1 0.1 0.2 0.2]]
[0.1 0.1 0.2 0.2]
Normalizing targets
done
WARNING: Logging before InitGoogleLogging() is written to STDERR
I0301 13:14:25.135601 26256 solver.cpp:48] Initializing solver from parameters:
train_net: "models/mitos/VGG_CNN_M_1024_Scale/faster_rcnn_end2end/train.prototxt"
base_lr: 0.001
display: 20
lr_policy: "step"
gamma: 0.1
momentum: 0.9
weight_decay: 0.0005
stepsize: 50000
snapshot: 0
snapshot_prefix: "vgg_cnn_m_1024_faster_rcnn"
average_loss: 100
I0301 13:14:25.135644 26256 solver.cpp:81] Creating training net from train_net file: models/mitos/VGG_CNN_                             M_1024_Scale/faster_rcnn_end2end/train.prototxt
I0301 13:14:25.136242 26256 net.cpp:49] Initializing net from parameters:
name: "VGG_CNN_M_1024"
state {
  phase: TRAIN
}
layer {
  name: "input-data"
  type: "Python"
  top: "data"
  top: "im_info"
  top: "gt_boxes"
  python_param {
    module: "roi_data_layer.layer"
    layer: "RoIDataLayer"
    param_str: "\'num_classes\': 2"
  }
}
layer {
  name: "conv1"
  type: "Convolution"
  bottom: "data"
  top: "conv1"
  param {
    lr_mult: 0
    decay_mult: 0
  }
  param {
    lr_mult: 0
    decay_mult: 0
  }
  convolution_param {
    num_output: 96
    kernel_size: 7
    stride: 2
  }
}
layer {
  name: "relu1"
  type: "ReLU"
  bottom: "conv1"
  top: "conv1"
}
layer {
  name: "norm1"
  type: "LRN"
  bottom: "conv1"
  top: "norm1"
  lrn_param {
    local_size: 5
    alpha: 0.0005
    beta: 0.75
    k: 2
  }
}
layer {
  name: "pool1"
  type: "Pooling"
  bottom: "norm1"
  top: "pool1"
  pooling_param {
    pool: MAX
    kernel_size: 3
    stride: 2
  }
}
layer {
  name: "conv2"
  type: "Convolution"
  bottom: "pool1"
  top: "conv2"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 256
    pad: 1
    kernel_size: 5
    stride: 2
  }
}
layer {
  name: "relu2"
  type: "ReLU"
  bottom: "conv2"
  top: "conv2"
}
layer {
  name: "norm2"
  type: "LRN"
  bottom: "conv2"
  top: "norm2"
  lrn_param {
    local_size: 5
    alpha: 0.0005
    beta: 0.75
    k: 2
  }
}
layer {
  name: "pool2"
  type: "Pooling"
  bottom: "norm2"
  top: "pool2"
  pooling_param {
    pool: MAX
    kernel_size: 3
    stride: 2
  }
}
layer {
  name: "conv3"
  type: "Convolution"
  bottom: "pool2"
  top: "conv3"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 512
    pad: 1
    kernel_size: 3
  }
}
layer {
  name: "relu3"
  type: "ReLU"
  bottom: "conv3"
  top: "conv3"
}
layer {
  name: "conv4"
  type: "Convolution"
  bottom: "conv3"
  top: "conv4"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 512
    pad: 1
    kernel_size: 3
  }
}
layer {
  name: "relu4"
  type: "ReLU"
  bottom: "conv4"
  top: "conv4"
}
layer {
  name: "conv5"
  type: "Convolution"
  bottom: "conv4"
  top: "conv5"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 512
    pad: 1
    kernel_size: 3
  }
}
layer {
  name: "relu5"
  type: "ReLU"
  bottom: "conv5"
  top: "conv5"
}
layer {
  name: "rpn_conv/3x3"
  type: "Convolution"
  bottom: "conv5"
  top: "rpn/output"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 256
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "rpn_relu/3x3"
  type: "ReLU"
  bottom: "rpn/output"
  top: "rpn/output"
}
layer {
  name: "rpn_cls_score"
  type: "Convolution"
  bottom: "rpn/output"
  top: "rpn_cls_score"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 18
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "rpn_bbox_pred"
  type: "Convolution"
  bottom: "rpn/output"
  top: "rpn_bbox_pred"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  convolution_param {
    num_output: 36
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "gaussian"
      std: 0.01
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "rpn_cls_score_reshape"
  type: "Reshape"
  bottom: "rpn_cls_score"
  top: "rpn_cls_score_reshape"
  reshape_param {
    shape {
      dim: 0
      dim: 2
      dim: -1
      dim: 0
    }
  }
}
layer {
  name: "rpn-data"
  type: "Python"
  bottom: "rpn_cls_score"
  bottom: "gt_boxes"
  bottom: "im_info"
  bottom: "data"
  top: "rpn_labels"
  top: "rpn_bbox_targets"
  top: "rpn_bbox_inside_weights"
  top: "rpn_bbox_outside_weights"
  python_param {
    module: "rpnS.anchor_target_layer"
    layer: "AnchorTargetLayer"
    param_str: "\'feat_stride\': 16"
  }
}
layer {
  name: "rpn_loss_cls"
  type: "SoftmaxWithLoss"
  bottom: "rpn_cls_score_reshape"
  bottom: "rpn_labels"
  top: "rpn_cls_loss"
  loss_weight: 1
  propagate_down: true
  propagate_down: false
  loss_param {
    ignore_label: -1
    normalize: true
  }
}
layer {
  name: "rpn_loss_bbox"
  type: "SmoothL1Loss"
  bottom: "rpn_bbox_pred"
  bottom: "rpn_bbox_targets"
  bottom: "rpn_bbox_inside_weights"
  bottom: "rpn_bbox_outside_weights"
  top: "rpn_loss_bbox"
  loss_weight: 1
  smooth_l1_loss_param {
    sigma: 3
  }
}
layer {
  name: "rpn_cls_prob"
  type: "Softmax"
  bottom: "rpn_cls_score_reshape"
  top: "rpn_cls_prob"
}
layer {
  name: "rpn_cls_prob_reshape"
  type: "Reshape"
  bottom: "rpn_cls_prob"
  top: "rpn_cls_prob_reshape"
  reshape_param {
    shape {
      dim: 0
      dim: 18
      dim: -1
      dim: 0
    }
  }
}
layer {
  name: "proposal"
  type: "Python"
  bottom: "rpn_cls_prob_reshape"
  bottom: "rpn_bbox_pred"
  bottom: "im_info"
  top: "rpn_rois"
  python_param {
    module: "rpnS.proposal_layer"
    layer: "ProposalLayer"
    param_str: "\'feat_stride\': 16"
  }
}
layer {
  name: "roi-data"
  type: "Python"
  bottom: "rpn_rois"
  bottom: "gt_boxes"
  top: "rois"
  top: "labels"
  top: "bbox_targets"
  top: "bbox_inside_weights"
  top: "bbox_outside_weights"
  python_param {
    module: "rpnS.proposal_target_layer"
    layer: "ProposalTargetLayer"
    param_str: "\'num_classes\': 2"
  }
}
layer {
  name: "roi_pool5"
  type: "ROIPooling"
  bottom: "conv5"
  bottom: "rois"
  top: "pool5"
  roi_pooling_param {
    pooled_h: 6
    pooled_w: 6
    spatial_scale: 0.0625
  }
}
layer {
  name: "fc6"
  type: "InnerProduct"
  bottom: "pool5"
  top: "fc6"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  inner_product_param {
    num_output: 4096
  }
}
layer {
  name: "relu6"
  type: "ReLU"
  bottom: "fc6"
  top: "fc6"
}
layer {
  name: "drop6"
  type: "Dropout"
  bottom: "fc6"
  top: "fc6"
  dropout_param {
    dropout_ratio: 0.5
  }
}
layer {
  name: "fc7"
  type: "InnerProduct"
  bottom: "fc6"
  top: "fc7"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  inner_product_param {
    num_output: 1024
  }
}
layer {
  name: "relu7"
  type: "ReLU"
  bottom: "fc7"
  top: "fc7"
}
layer {
  name: "drop7"
  type: "Dropout"
  bottom: "fc7"
  top: "fc7"
  dropout_param {
    dropout_ratio: 0.5
  }
}
layer {
  name: "cls_score"
  type: "InnerProduct"
  bottom: "fc7"
  top: "cls_score"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  inner_product_param {
    num_output: 2
    weight_filler {
      type: "gaussian"
      std: 0.01
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "bbox_pred"
  type: "InnerProduct"
  bottom: "fc7"
  top: "bbox_pred"
  param {
    lr_mult: 1
  }
  param {
    lr_mult: 2
  }
  inner_product_param {
    num_output: 8
    weight_filler {
      type: "gaussian"
      std: 0.001
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "loss_cls"
  type: "SoftmaxWithLoss"
  bottom: "cls_score"
  bottom: "labels"
  top: "loss_cls"
  loss_weight: 1
  propagate_down: true
  propagate_down: false
}
layer {
  name: "loss_bbox"
  type: "SmoothL1Loss"
  bottom: "bbox_pred"
  bottom: "bbox_targets"
  bottom: "bbox_inside_weights"
  bottom: "bbox_outside_weights"
  top: "loss_bbox"
  loss_weight: 1
}
I0301 13:14:25.136497 26256 layer_factory.hpp:77] Creating layer input-data
I0301 13:14:25.167214 26256 net.cpp:106] Creating Layer input-data
I0301 13:14:25.167241 26256 net.cpp:411] input-data -> data
I0301 13:14:25.167254 26256 net.cpp:411] input-data -> im_info
I0301 13:14:25.167261 26256 net.cpp:411] input-data -> gt_boxes
RoiDataLayer: name_to_top: {'gt_boxes': 2, 'data': 0, 'im_info': 1}
I0301 13:14:25.266619 26256 net.cpp:150] Setting up input-data
I0301 13:14:25.266654 26256 net.cpp:157] Top shape: 1 3 1024 1024 (3145728)
I0301 13:14:25.266659 26256 net.cpp:157] Top shape: 1 3 (3)
I0301 13:14:25.266662 26256 net.cpp:157] Top shape: 1 4 (4)
I0301 13:14:25.266665 26256 net.cpp:165] Memory required for data: 12582940
I0301 13:14:25.266674 26256 layer_factory.hpp:77] Creating layer data_input-data_0_split
I0301 13:14:25.266690 26256 net.cpp:106] Creating Layer data_input-data_0_split
I0301 13:14:25.266695 26256 net.cpp:454] data_input-data_0_split <- data
I0301 13:14:25.266705 26256 net.cpp:411] data_input-data_0_split -> data_input-data_0_split_0
I0301 13:14:25.266716 26256 net.cpp:411] data_input-data_0_split -> data_input-data_0_split_1
I0301 13:14:25.266777 26256 net.cpp:150] Setting up data_input-data_0_split
I0301 13:14:25.266782 26256 net.cpp:157] Top shape: 1 3 1024 1024 (3145728)
I0301 13:14:25.266788 26256 net.cpp:157] Top shape: 1 3 1024 1024 (3145728)
I0301 13:14:25.266790 26256 net.cpp:165] Memory required for data: 37748764
I0301 13:14:25.266793 26256 layer_factory.hpp:77] Creating layer im_info_input-data_1_split
I0301 13:14:25.266799 26256 net.cpp:106] Creating Layer im_info_input-data_1_split
I0301 13:14:25.266803 26256 net.cpp:454] im_info_input-data_1_split <- im_info
I0301 13:14:25.266808 26256 net.cpp:411] im_info_input-data_1_split -> im_info_input-data_1_split_0
I0301 13:14:25.266814 26256 net.cpp:411] im_info_input-data_1_split -> im_info_input-data_1_split_1
I0301 13:14:25.266862 26256 net.cpp:150] Setting up im_info_input-data_1_split
I0301 13:14:25.266867 26256 net.cpp:157] Top shape: 1 3 (3)
I0301 13:14:25.266870 26256 net.cpp:157] Top shape: 1 3 (3)
I0301 13:14:25.266873 26256 net.cpp:165] Memory required for data: 37748788
I0301 13:14:25.266876 26256 layer_factory.hpp:77] Creating layer gt_boxes_input-data_2_split
I0301 13:14:25.266880 26256 net.cpp:106] Creating Layer gt_boxes_input-data_2_split
I0301 13:14:25.266883 26256 net.cpp:454] gt_boxes_input-data_2_split <- gt_boxes
I0301 13:14:25.266888 26256 net.cpp:411] gt_boxes_input-data_2_split -> gt_boxes_input-data_2_split_0
I0301 13:14:25.266894 26256 net.cpp:411] gt_boxes_input-data_2_split -> gt_boxes_input-data_2_split_1
I0301 13:14:25.266942 26256 net.cpp:150] Setting up gt_boxes_input-data_2_split
I0301 13:14:25.266947 26256 net.cpp:157] Top shape: 1 4 (4)
I0301 13:14:25.266952 26256 net.cpp:157] Top shape: 1 4 (4)
I0301 13:14:25.266954 26256 net.cpp:165] Memory required for data: 37748820
I0301 13:14:25.266957 26256 layer_factory.hpp:77] Creating layer conv1
I0301 13:14:25.266969 26256 net.cpp:106] Creating Layer conv1
I0301 13:14:25.266973 26256 net.cpp:454] conv1 <- data_input-data_0_split_0
I0301 13:14:25.266978 26256 net.cpp:411] conv1 -> conv1
I0301 13:14:28.179129 26256 net.cpp:150] Setting up conv1
I0301 13:14:28.179168 26256 net.cpp:157] Top shape: 1 96 509 509 (24871776)
I0301 13:14:28.179172 26256 net.cpp:165] Memory required for data: 137235924
I0301 13:14:28.179198 26256 layer_factory.hpp:77] Creating layer relu1
I0301 13:14:28.179214 26256 net.cpp:106] Creating Layer relu1
I0301 13:14:28.179224 26256 net.cpp:454] relu1 <- conv1
I0301 13:14:28.179234 26256 net.cpp:397] relu1 -> conv1 (in-place)
I0301 13:14:28.186056 26256 net.cpp:150] Setting up relu1
I0301 13:14:28.186096 26256 net.cpp:157] Top shape: 1 96 509 509 (24871776)
I0301 13:14:28.186101 26256 net.cpp:165] Memory required for data: 236723028
I0301 13:14:28.186111 26256 layer_factory.hpp:77] Creating layer norm1
I0301 13:14:28.186134 26256 net.cpp:106] Creating Layer norm1
I0301 13:14:28.186143 26256 net.cpp:454] norm1 <- conv1
I0301 13:14:28.186157 26256 net.cpp:411] norm1 -> norm1
I0301 13:14:28.186389 26256 net.cpp:150] Setting up norm1
I0301 13:14:28.186403 26256 net.cpp:157] Top shape: 1 96 509 509 (24871776)
I0301 13:14:28.186408 26256 net.cpp:165] Memory required for data: 336210132
I0301 13:14:28.186415 26256 layer_factory.hpp:77] Creating layer pool1
I0301 13:14:28.186440 26256 net.cpp:106] Creating Layer pool1
I0301 13:14:28.186445 26256 net.cpp:454] pool1 <- norm1
I0301 13:14:28.186455 26256 net.cpp:411] pool1 -> pool1
I0301 13:14:28.186538 26256 net.cpp:150] Setting up pool1
I0301 13:14:28.186547 26256 net.cpp:157] Top shape: 1 96 254 254 (6193536)
I0301 13:14:28.186551 26256 net.cpp:165] Memory required for data: 360984276
I0301 13:14:28.186556 26256 layer_factory.hpp:77] Creating layer conv2
I0301 13:14:28.186581 26256 net.cpp:106] Creating Layer conv2
I0301 13:14:28.186587 26256 net.cpp:454] conv2 <- pool1
I0301 13:14:28.186596 26256 net.cpp:411] conv2 -> conv2
I0301 13:14:28.201721 26256 net.cpp:150] Setting up conv2
I0301 13:14:28.201782 26256 net.cpp:157] Top shape: 1 256 126 126 (4064256)
I0301 13:14:28.201789 26256 net.cpp:165] Memory required for data: 377241300
I0301 13:14:28.201815 26256 layer_factory.hpp:77] Creating layer relu2
I0301 13:14:28.201835 26256 net.cpp:106] Creating Layer relu2
I0301 13:14:28.201843 26256 net.cpp:454] relu2 <- conv2
I0301 13:14:28.201853 26256 net.cpp:397] relu2 -> conv2 (in-place)
I0301 13:14:28.204072 26256 net.cpp:150] Setting up relu2
I0301 13:14:28.204104 26256 net.cpp:157] Top shape: 1 256 126 126 (4064256)
I0301 13:14:28.204108 26256 net.cpp:165] Memory required for data: 393498324
I0301 13:14:28.204116 26256 layer_factory.hpp:77] Creating layer norm2
I0301 13:14:28.204135 26256 net.cpp:106] Creating Layer norm2
I0301 13:14:28.204144 26256 net.cpp:454] norm2 <- conv2
I0301 13:14:28.204154 26256 net.cpp:411] norm2 -> norm2
I0301 13:14:28.206008 26256 net.cpp:150] Setting up norm2
I0301 13:14:28.206019 26256 net.cpp:157] Top shape: 1 256 126 126 (4064256)
I0301 13:14:28.206022 26256 net.cpp:165] Memory required for data: 409755348
I0301 13:14:28.206027 26256 layer_factory.hpp:77] Creating layer pool2
I0301 13:14:28.206035 26256 net.cpp:106] Creating Layer pool2
I0301 13:14:28.206043 26256 net.cpp:454] pool2 <- norm2
I0301 13:14:28.206048 26256 net.cpp:411] pool2 -> pool2
I0301 13:14:28.206109 26256 net.cpp:150] Setting up pool2
I0301 13:14:28.206118 26256 net.cpp:157] Top shape: 1 256 63 63 (1016064)
I0301 13:14:28.206122 26256 net.cpp:165] Memory required for data: 413819604
I0301 13:14:28.206125 26256 layer_factory.hpp:77] Creating layer conv3
I0301 13:14:28.206137 26256 net.cpp:106] Creating Layer conv3
I0301 13:14:28.206141 26256 net.cpp:454] conv3 <- pool2
I0301 13:14:28.206149 26256 net.cpp:411] conv3 -> conv3
I0301 13:14:28.231119 26256 net.cpp:150] Setting up conv3
I0301 13:14:28.231153 26256 net.cpp:157] Top shape: 1 512 63 63 (2032128)
I0301 13:14:28.231156 26256 net.cpp:165] Memory required for data: 421948116
I0301 13:14:28.231181 26256 layer_factory.hpp:77] Creating layer relu3
I0301 13:14:28.231196 26256 net.cpp:106] Creating Layer relu3
I0301 13:14:28.231221 26256 net.cpp:454] relu3 <- conv3
I0301 13:14:28.231230 26256 net.cpp:397] relu3 -> conv3 (in-place)
I0301 13:14:28.232005 26256 net.cpp:150] Setting up relu3
I0301 13:14:28.232023 26256 net.cpp:157] Top shape: 1 512 63 63 (2032128)
I0301 13:14:28.232028 26256 net.cpp:165] Memory required for data: 430076628
I0301 13:14:28.232034 26256 layer_factory.hpp:77] Creating layer conv4
I0301 13:14:28.232059 26256 net.cpp:106] Creating Layer conv4
I0301 13:14:28.232069 26256 net.cpp:454] conv4 <- conv3
I0301 13:14:28.232080 26256 net.cpp:411] conv4 -> conv4
I0301 13:14:28.251152 26256 net.cpp:150] Setting up conv4
I0301 13:14:28.251204 26256 net.cpp:157] Top shape: 1 512 63 63 (2032128)
I0301 13:14:28.251209 26256 net.cpp:165] Memory required for data: 438205140
I0301 13:14:28.251226 26256 layer_factory.hpp:77] Creating layer relu4
I0301 13:14:28.251247 26256 net.cpp:106] Creating Layer relu4
I0301 13:14:28.251255 26256 net.cpp:454] relu4 <- conv4
I0301 13:14:28.251266 26256 net.cpp:397] relu4 -> conv4 (in-place)
I0301 13:14:28.253309 26256 net.cpp:150] Setting up relu4
I0301 13:14:28.253324 26256 net.cpp:157] Top shape: 1 512 63 63 (2032128)
I0301 13:14:28.253329 26256 net.cpp:165] Memory required for data: 446333652
I0301 13:14:28.253334 26256 layer_factory.hpp:77] Creating layer conv5
I0301 13:14:28.253378 26256 net.cpp:106] Creating Layer conv5
I0301 13:14:28.253386 26256 net.cpp:454] conv5 <- conv4
I0301 13:14:28.253396 26256 net.cpp:411] conv5 -> conv5
I0301 13:14:28.271078 26256 net.cpp:150] Setting up conv5
I0301 13:14:28.271123 26256 net.cpp:157] Top shape: 1 512 63 63 (2032128)
I0301 13:14:28.271129 26256 net.cpp:165] Memory required for data: 454462164
I0301 13:14:28.271160 26256 layer_factory.hpp:77] Creating layer relu5
I0301 13:14:28.271179 26256 net.cpp:106] Creating Layer relu5
I0301 13:14:28.271188 26256 net.cpp:454] relu5 <- conv5
I0301 13:14:28.271203 26256 net.cpp:397] relu5 -> conv5 (in-place)
I0301 13:14:28.274314 26256 net.cpp:150] Setting up relu5
I0301 13:14:28.274358 26256 net.cpp:157] Top shape: 1 512 63 63 (2032128)
I0301 13:14:28.274363 26256 net.cpp:165] Memory required for data: 462590676
I0301 13:14:28.274374 26256 layer_factory.hpp:77] Creating layer conv5_relu5_0_split
I0301 13:14:28.274394 26256 net.cpp:106] Creating Layer conv5_relu5_0_split
I0301 13:14:28.274402 26256 net.cpp:454] conv5_relu5_0_split <- conv5
I0301 13:14:28.274416 26256 net.cpp:411] conv5_relu5_0_split -> conv5_relu5_0_split_0
I0301 13:14:28.274430 26256 net.cpp:411] conv5_relu5_0_split -> conv5_relu5_0_split_1
I0301 13:14:28.274574 26256 net.cpp:150] Setting up conv5_relu5_0_split
I0301 13:14:28.274587 26256 net.cpp:157] Top shape: 1 512 63 63 (2032128)
I0301 13:14:28.274595 26256 net.cpp:157] Top shape: 1 512 63 63 (2032128)
I0301 13:14:28.274600 26256 net.cpp:165] Memory required for data: 478847700
I0301 13:14:28.274605 26256 layer_factory.hpp:77] Creating layer rpn_conv/3x3
I0301 13:14:28.274639 26256 net.cpp:106] Creating Layer rpn_conv/3x3
I0301 13:14:28.274657 26256 net.cpp:454] rpn_conv/3x3 <- conv5_relu5_0_split_0
I0301 13:14:28.274674 26256 net.cpp:411] rpn_conv/3x3 -> rpn/output
I0301 13:14:28.311064 26256 net.cpp:150] Setting up rpn_conv/3x3
I0301 13:14:28.311117 26256 net.cpp:157] Top shape: 1 256 63 63 (1016064)
I0301 13:14:28.311121 26256 net.cpp:165] Memory required for data: 482911956
I0301 13:14:28.311136 26256 layer_factory.hpp:77] Creating layer rpn_relu/3x3
I0301 13:14:28.311149 26256 net.cpp:106] Creating Layer rpn_relu/3x3
I0301 13:14:28.311156 26256 net.cpp:454] rpn_relu/3x3 <- rpn/output
I0301 13:14:28.311165 26256 net.cpp:397] rpn_relu/3x3 -> rpn/output (in-place)
I0301 13:14:28.312248 26256 net.cpp:150] Setting up rpn_relu/3x3
I0301 13:14:28.312260 26256 net.cpp:157] Top shape: 1 256 63 63 (1016064)
I0301 13:14:28.312263 26256 net.cpp:165] Memory required for data: 486976212
I0301 13:14:28.312268 26256 layer_factory.hpp:77] Creating layer rpn/output_rpn_relu/3x3_0_split
I0301 13:14:28.312281 26256 net.cpp:106] Creating Layer rpn/output_rpn_relu/3x3_0_split
I0301 13:14:28.312284 26256 net.cpp:454] rpn/output_rpn_relu/3x3_0_split <- rpn/output
I0301 13:14:28.312292 26256 net.cpp:411] rpn/output_rpn_relu/3x3_0_split -> rpn/output_rpn_relu/3x3_0_split                             _0
I0301 13:14:28.312300 26256 net.cpp:411] rpn/output_rpn_relu/3x3_0_split -> rpn/output_rpn_relu/3x3_0_split                             _1
I0301 13:14:28.312366 26256 net.cpp:150] Setting up rpn/output_rpn_relu/3x3_0_split
I0301 13:14:28.312371 26256 net.cpp:157] Top shape: 1 256 63 63 (1016064)
I0301 13:14:28.312374 26256 net.cpp:157] Top shape: 1 256 63 63 (1016064)
I0301 13:14:28.312376 26256 net.cpp:165] Memory required for data: 495104724
I0301 13:14:28.312381 26256 layer_factory.hpp:77] Creating layer rpn_cls_score
I0301 13:14:28.312398 26256 net.cpp:106] Creating Layer rpn_cls_score
I0301 13:14:28.312402 26256 net.cpp:454] rpn_cls_score <- rpn/output_rpn_relu/3x3_0_split_0
I0301 13:14:28.312410 26256 net.cpp:411] rpn_cls_score -> rpn_cls_score
I0301 13:14:28.319301 26256 net.cpp:150] Setting up rpn_cls_score
I0301 13:14:28.319325 26256 net.cpp:157] Top shape: 1 18 63 63 (71442)
I0301 13:14:28.319329 26256 net.cpp:165] Memory required for data: 495390492
I0301 13:14:28.319339 26256 layer_factory.hpp:77] Creating layer rpn_cls_score_rpn_cls_score_0_split
I0301 13:14:28.319353 26256 net.cpp:106] Creating Layer rpn_cls_score_rpn_cls_score_0_split
I0301 13:14:28.319370 26256 net.cpp:454] rpn_cls_score_rpn_cls_score_0_split <- rpn_cls_score
I0301 13:14:28.319380 26256 net.cpp:411] rpn_cls_score_rpn_cls_score_0_split -> rpn_cls_score_rpn_cls_score                             _0_split_0
I0301 13:14:28.319396 26256 net.cpp:411] rpn_cls_score_rpn_cls_score_0_split -> rpn_cls_score_rpn_cls_score                             _0_split_1
I0301 13:14:28.319473 26256 net.cpp:150] Setting up rpn_cls_score_rpn_cls_score_0_split
I0301 13:14:28.319484 26256 net.cpp:157] Top shape: 1 18 63 63 (71442)
I0301 13:14:28.319495 26256 net.cpp:157] Top shape: 1 18 63 63 (71442)
I0301 13:14:28.319502 26256 net.cpp:165] Memory required for data: 495962028
I0301 13:14:28.319506 26256 layer_factory.hpp:77] Creating layer rpn_bbox_pred
I0301 13:14:28.319535 26256 net.cpp:106] Creating Layer rpn_bbox_pred
I0301 13:14:28.319542 26256 net.cpp:454] rpn_bbox_pred <- rpn/output_rpn_relu/3x3_0_split_1
I0301 13:14:28.319555 26256 net.cpp:411] rpn_bbox_pred -> rpn_bbox_pred
I0301 13:14:28.333879 26256 net.cpp:150] Setting up rpn_bbox_pred
I0301 13:14:28.333915 26256 net.cpp:157] Top shape: 1 36 63 63 (142884)
I0301 13:14:28.333918 26256 net.cpp:165] Memory required for data: 496533564
I0301 13:14:28.333948 26256 layer_factory.hpp:77] Creating layer rpn_bbox_pred_rpn_bbox_pred_0_split
I0301 13:14:28.334079 26256 net.cpp:106] Creating Layer rpn_bbox_pred_rpn_bbox_pred_0_split
I0301 13:14:28.334097 26256 net.cpp:454] rpn_bbox_pred_rpn_bbox_pred_0_split <- rpn_bbox_pred
I0301 13:14:28.334118 26256 net.cpp:411] rpn_bbox_pred_rpn_bbox_pred_0_split -> rpn_bbox_pred_rpn_bbox_pred                             _0_split_0
I0301 13:14:28.334141 26256 net.cpp:411] rpn_bbox_pred_rpn_bbox_pred_0_split -> rpn_bbox_pred_rpn_bbox_pred                             _0_split_1
I0301 13:14:28.334300 26256 net.cpp:150] Setting up rpn_bbox_pred_rpn_bbox_pred_0_split
I0301 13:14:28.334314 26256 net.cpp:157] Top shape: 1 36 63 63 (142884)
I0301 13:14:28.334318 26256 net.cpp:157] Top shape: 1 36 63 63 (142884)
I0301 13:14:28.334321 26256 net.cpp:165] Memory required for data: 497676636
I0301 13:14:28.334326 26256 layer_factory.hpp:77] Creating layer rpn_cls_score_reshape
I0301 13:14:28.334336 26256 net.cpp:106] Creating Layer rpn_cls_score_reshape
I0301 13:14:28.334342 26256 net.cpp:454] rpn_cls_score_reshape <- rpn_cls_score_rpn_cls_score_0_split_0
I0301 13:14:28.334353 26256 net.cpp:411] rpn_cls_score_reshape -> rpn_cls_score_reshape
I0301 13:14:28.334419 26256 net.cpp:150] Setting up rpn_cls_score_reshape
I0301 13:14:28.334425 26256 net.cpp:157] Top shape: 1 2 567 63 (71442)
I0301 13:14:28.334429 26256 net.cpp:165] Memory required for data: 497962404
I0301 13:14:28.334434 26256 layer_factory.hpp:77] Creating layer rpn_cls_score_reshape_rpn_cls_score_reshap                             e_0_split
I0301 13:14:28.334439 26256 net.cpp:106] Creating Layer rpn_cls_score_reshape_rpn_cls_score_reshape_0_split
I0301 13:14:28.334442 26256 net.cpp:454] rpn_cls_score_reshape_rpn_cls_score_reshape_0_split <- rpn_cls_sco                             re_reshape
I0301 13:14:28.334451 26256 net.cpp:411] rpn_cls_score_reshape_rpn_cls_score_reshape_0_split -> rpn_cls_sco                             re_reshape_rpn_cls_score_reshape_0_split_0
I0301 13:14:28.334457 26256 net.cpp:411] rpn_cls_score_reshape_rpn_cls_score_reshape_0_split -> rpn_cls_sco                             re_reshape_rpn_cls_score_reshape_0_split_1
I0301 13:14:28.334519 26256 net.cpp:150] Setting up rpn_cls_score_reshape_rpn_cls_score_reshape_0_split
I0301 13:14:28.334527 26256 net.cpp:157] Top shape: 1 2 567 63 (71442)
I0301 13:14:28.334530 26256 net.cpp:157] Top shape: 1 2 567 63 (71442)
I0301 13:14:28.334532 26256 net.cpp:165] Memory required for data: 498533940
I0301 13:14:28.334537 26256 layer_factory.hpp:77] Creating layer rpn-data
I0301 13:14:28.335705 26256 net.cpp:106] Creating Layer rpn-data
I0301 13:14:28.335736 26256 net.cpp:454] rpn-data <- rpn_cls_score_rpn_cls_score_0_split_1
I0301 13:14:28.335745 26256 net.cpp:454] rpn-data <- gt_boxes_input-data_2_split_0
I0301 13:14:28.335750 26256 net.cpp:454] rpn-data <- im_info_input-data_1_split_0
I0301 13:14:28.335755 26256 net.cpp:454] rpn-data <- data_input-data_0_split_1
I0301 13:14:28.335764 26256 net.cpp:411] rpn-data -> rpn_labels
I0301 13:14:28.335789 26256 net.cpp:411] rpn-data -> rpn_bbox_targets
I0301 13:14:28.335804 26256 net.cpp:411] rpn-data -> rpn_bbox_inside_weights
I0301 13:14:28.335816 26256 net.cpp:411] rpn-data -> rpn_bbox_outside_weights
I0301 13:14:28.337988 26256 net.cpp:150] Setting up rpn-data
I0301 13:14:28.338021 26256 net.cpp:157] Top shape: 1 1 567 63 (35721)
I0301 13:14:28.338027 26256 net.cpp:157] Top shape: 1 36 63 63 (142884)
I0301 13:14:28.338032 26256 net.cpp:157] Top shape: 1 36 63 63 (142884)
I0301 13:14:28.338037 26256 net.cpp:157] Top shape: 1 36 63 63 (142884)
I0301 13:14:28.338042 26256 net.cpp:165] Memory required for data: 500391432
I0301 13:14:28.338048 26256 layer_factory.hpp:77] Creating layer rpn_loss_cls
I0301 13:14:28.338061 26256 net.cpp:106] Creating Layer rpn_loss_cls
I0301 13:14:28.338071 26256 net.cpp:454] rpn_loss_cls <- rpn_cls_score_reshape_rpn_cls_score_reshape_0_spli                             t_0
I0301 13:14:28.338083 26256 net.cpp:454] rpn_loss_cls <- rpn_labels
I0301 13:14:28.338095 26256 net.cpp:411] rpn_loss_cls -> rpn_cls_loss
I0301 13:14:28.338114 26256 layer_factory.hpp:77] Creating layer rpn_loss_cls
I0301 13:14:28.340728 26256 net.cpp:150] Setting up rpn_loss_cls
I0301 13:14:28.340754 26256 net.cpp:157] Top shape: (1)
I0301 13:14:28.340759 26256 net.cpp:160]     with loss weight 1
I0301 13:14:28.340785 26256 net.cpp:165] Memory required for data: 500391436
I0301 13:14:28.340790 26256 layer_factory.hpp:77] Creating layer rpn_loss_bbox
I0301 13:14:28.340812 26256 net.cpp:106] Creating Layer rpn_loss_bbox
I0301 13:14:28.340821 26256 net.cpp:454] rpn_loss_bbox <- rpn_bbox_pred_rpn_bbox_pred_0_split_0
I0301 13:14:28.340833 26256 net.cpp:454] rpn_loss_bbox <- rpn_bbox_targets
I0301 13:14:28.340838 26256 net.cpp:454] rpn_loss_bbox <- rpn_bbox_inside_weights
I0301 13:14:28.340844 26256 net.cpp:454] rpn_loss_bbox <- rpn_bbox_outside_weights
I0301 13:14:28.340850 26256 net.cpp:411] rpn_loss_bbox -> rpn_loss_bbox
I0301 13:14:28.344079 26256 net.cpp:150] Setting up rpn_loss_bbox
I0301 13:14:28.344125 26256 net.cpp:157] Top shape: (1)
I0301 13:14:28.344128 26256 net.cpp:160]     with loss weight 1
I0301 13:14:28.344151 26256 net.cpp:165] Memory required for data: 500391440
I0301 13:14:28.344159 26256 layer_factory.hpp:77] Creating layer rpn_cls_prob
I0301 13:14:28.344190 26256 net.cpp:106] Creating Layer rpn_cls_prob
I0301 13:14:28.344205 26256 net.cpp:454] rpn_cls_prob <- rpn_cls_score_reshape_rpn_cls_score_reshape_0_spli                             t_1
I0301 13:14:28.344223 26256 net.cpp:411] rpn_cls_prob -> rpn_cls_prob
I0301 13:14:28.344583 26256 net.cpp:150] Setting up rpn_cls_prob
I0301 13:14:28.344596 26256 net.cpp:157] Top shape: 1 2 567 63 (71442)
I0301 13:14:28.344601 26256 net.cpp:165] Memory required for data: 500677208
I0301 13:14:28.344606 26256 layer_factory.hpp:77] Creating layer rpn_cls_prob_reshape
I0301 13:14:28.344641 26256 net.cpp:106] Creating Layer rpn_cls_prob_reshape
I0301 13:14:28.344648 26256 net.cpp:454] rpn_cls_prob_reshape <- rpn_cls_prob
I0301 13:14:28.344671 26256 net.cpp:411] rpn_cls_prob_reshape -> rpn_cls_prob_reshape
I0301 13:14:28.344733 26256 net.cpp:150] Setting up rpn_cls_prob_reshape
I0301 13:14:28.344743 26256 net.cpp:157] Top shape: 1 18 63 63 (71442)
I0301 13:14:28.344746 26256 net.cpp:165] Memory required for data: 500962976
I0301 13:14:28.344751 26256 layer_factory.hpp:77] Creating layer proposal
I0301 13:14:28.364584 26256 net.cpp:106] Creating Layer proposal
I0301 13:14:28.364621 26256 net.cpp:454] proposal <- rpn_cls_prob_reshape
I0301 13:14:28.364634 26256 net.cpp:454] proposal <- rpn_bbox_pred_rpn_bbox_pred_0_split_1
I0301 13:14:28.364640 26256 net.cpp:454] proposal <- im_info_input-data_1_split_1
I0301 13:14:28.364668 26256 net.cpp:411] proposal -> rpn_rois
I0301 13:14:28.367686 26256 net.cpp:150] Setting up proposal
I0301 13:14:28.367710 26256 net.cpp:157] Top shape: 1 5 (5)
I0301 13:14:28.367717 26256 net.cpp:165] Memory required for data: 500962996
I0301 13:14:28.367725 26256 layer_factory.hpp:77] Creating layer roi-data
I0301 13:14:28.370525 26256 net.cpp:106] Creating Layer roi-data
I0301 13:14:28.370553 26256 net.cpp:454] roi-data <- rpn_rois
I0301 13:14:28.370571 26256 net.cpp:454] roi-data <- gt_boxes_input-data_2_split_1
I0301 13:14:28.370594 26256 net.cpp:411] roi-data -> rois
I0301 13:14:28.370631 26256 net.cpp:411] roi-data -> labels
I0301 13:14:28.370654 26256 net.cpp:411] roi-data -> bbox_targets
I0301 13:14:28.370672 26256 net.cpp:411] roi-data -> bbox_inside_weights
I0301 13:14:28.370688 26256 net.cpp:411] roi-data -> bbox_outside_weights
I0301 13:14:28.371800 26256 net.cpp:150] Setting up roi-data
I0301 13:14:28.371832 26256 net.cpp:157] Top shape: 1 5 (5)
I0301 13:14:28.371850 26256 net.cpp:157] Top shape: 1 1 (1)
I0301 13:14:28.371865 26256 net.cpp:157] Top shape: 1 8 (8)
I0301 13:14:28.371876 26256 net.cpp:157] Top shape: 1 8 (8)
I0301 13:14:28.371898 26256 net.cpp:157] Top shape: 1 8 (8)
I0301 13:14:28.371904 26256 net.cpp:165] Memory required for data: 500963116
I0301 13:14:28.371914 26256 layer_factory.hpp:77] Creating layer roi_pool5
I0301 13:14:28.371927 26256 net.cpp:106] Creating Layer roi_pool5
I0301 13:14:28.371934 26256 net.cpp:454] roi_pool5 <- conv5_relu5_0_split_1
I0301 13:14:28.371942 26256 net.cpp:454] roi_pool5 <- rois
I0301 13:14:28.371953 26256 net.cpp:411] roi_pool5 -> pool5
I0301 13:14:28.371963 26256 roi_pooling_layer.cpp:30] Spatial scale: 0.0625
I0301 13:14:28.372123 26256 net.cpp:150] Setting up roi_pool5
I0301 13:14:28.372145 26256 net.cpp:157] Top shape: 1 512 6 6 (18432)
I0301 13:14:28.372159 26256 net.cpp:165] Memory required for data: 501036844
I0301 13:14:28.372172 26256 layer_factory.hpp:77] Creating layer fc6
I0301 13:14:28.372196 26256 net.cpp:106] Creating Layer fc6
I0301 13:14:28.372205 26256 net.cpp:454] fc6 <- pool5
I0301 13:14:28.372220 26256 net.cpp:411] fc6 -> fc6
I0301 13:14:28.715776 26256 net.cpp:150] Setting up fc6
I0301 13:14:28.715816 26256 net.cpp:157] Top shape: 1 4096 (4096)
I0301 13:14:28.715819 26256 net.cpp:165] Memory required for data: 501053228
I0301 13:14:28.715842 26256 layer_factory.hpp:77] Creating layer relu6
I0301 13:14:28.715855 26256 net.cpp:106] Creating Layer relu6
I0301 13:14:28.715860 26256 net.cpp:454] relu6 <- fc6
I0301 13:14:28.715869 26256 net.cpp:397] relu6 -> fc6 (in-place)
I0301 13:14:28.716799 26256 net.cpp:150] Setting up relu6
I0301 13:14:28.716812 26256 net.cpp:157] Top shape: 1 4096 (4096)
I0301 13:14:28.716815 26256 net.cpp:165] Memory required for data: 501069612
I0301 13:14:28.716820 26256 layer_factory.hpp:77] Creating layer drop6
I0301 13:14:28.716848 26256 net.cpp:106] Creating Layer drop6
I0301 13:14:28.716854 26256 net.cpp:454] drop6 <- fc6
I0301 13:14:28.716861 26256 net.cpp:397] drop6 -> fc6 (in-place)
I0301 13:14:28.716910 26256 net.cpp:150] Setting up drop6
I0301 13:14:28.716918 26256 net.cpp:157] Top shape: 1 4096 (4096)
I0301 13:14:28.716920 26256 net.cpp:165] Memory required for data: 501085996
I0301 13:14:28.716923 26256 layer_factory.hpp:77] Creating layer fc7
I0301 13:14:28.716934 26256 net.cpp:106] Creating Layer fc7
I0301 13:14:28.716938 26256 net.cpp:454] fc7 <- fc6
I0301 13:14:28.716943 26256 net.cpp:411] fc7 -> fc7
I0301 13:14:28.732230 26256 net.cpp:150] Setting up fc7
I0301 13:14:28.732285 26256 net.cpp:157] Top shape: 1 1024 (1024)
I0301 13:14:28.732288 26256 net.cpp:165] Memory required for data: 501090092
I0301 13:14:28.732302 26256 layer_factory.hpp:77] Creating layer relu7
I0301 13:14:28.732316 26256 net.cpp:106] Creating Layer relu7
I0301 13:14:28.732336 26256 net.cpp:454] relu7 <- fc7
I0301 13:14:28.732347 26256 net.cpp:397] relu7 -> fc7 (in-place)
I0301 13:14:28.732522 26256 net.cpp:150] Setting up relu7
I0301 13:14:28.732535 26256 net.cpp:157] Top shape: 1 1024 (1024)
I0301 13:14:28.732539 26256 net.cpp:165] Memory required for data: 501094188
I0301 13:14:28.732555 26256 layer_factory.hpp:77] Creating layer drop7
I0301 13:14:28.732566 26256 net.cpp:106] Creating Layer drop7
I0301 13:14:28.732571 26256 net.cpp:454] drop7 <- fc7
I0301 13:14:28.732581 26256 net.cpp:397] drop7 -> fc7 (in-place)
I0301 13:14:28.732635 26256 net.cpp:150] Setting up drop7
I0301 13:14:28.732645 26256 net.cpp:157] Top shape: 1 1024 (1024)
I0301 13:14:28.732648 26256 net.cpp:165] Memory required for data: 501098284
I0301 13:14:28.732653 26256 layer_factory.hpp:77] Creating layer fc7_drop7_0_split
I0301 13:14:28.732663 26256 net.cpp:106] Creating Layer fc7_drop7_0_split
I0301 13:14:28.732668 26256 net.cpp:454] fc7_drop7_0_split <- fc7
I0301 13:14:28.732681 26256 net.cpp:411] fc7_drop7_0_split -> fc7_drop7_0_split_0
I0301 13:14:28.732692 26256 net.cpp:411] fc7_drop7_0_split -> fc7_drop7_0_split_1
I0301 13:14:28.732787 26256 net.cpp:150] Setting up fc7_drop7_0_split
I0301 13:14:28.732796 26256 net.cpp:157] Top shape: 1 1024 (1024)
I0301 13:14:28.732802 26256 net.cpp:157] Top shape: 1 1024 (1024)
I0301 13:14:28.732806 26256 net.cpp:165] Memory required for data: 501106476
I0301 13:14:28.732812 26256 layer_factory.hpp:77] Creating layer cls_score
I0301 13:14:28.732827 26256 net.cpp:106] Creating Layer cls_score
I0301 13:14:28.732842 26256 net.cpp:454] cls_score <- fc7_drop7_0_split_0
I0301 13:14:28.732852 26256 net.cpp:411] cls_score -> cls_score
I0301 13:14:28.733062 26256 net.cpp:150] Setting up cls_score
I0301 13:14:28.733072 26256 net.cpp:157] Top shape: 1 2 (2)
I0301 13:14:28.733075 26256 net.cpp:165] Memory required for data: 501106484
I0301 13:14:28.733085 26256 layer_factory.hpp:77] Creating layer bbox_pred
I0301 13:14:28.733099 26256 net.cpp:106] Creating Layer bbox_pred
I0301 13:14:28.733105 26256 net.cpp:454] bbox_pred <- fc7_drop7_0_split_1
I0301 13:14:28.733114 26256 net.cpp:411] bbox_pred -> bbox_pred
I0301 13:14:28.733412 26256 net.cpp:150] Setting up bbox_pred
I0301 13:14:28.733423 26256 net.cpp:157] Top shape: 1 8 (8)
I0301 13:14:28.733428 26256 net.cpp:165] Memory required for data: 501106516
I0301 13:14:28.733436 26256 layer_factory.hpp:77] Creating layer loss_cls
I0301 13:14:28.733448 26256 net.cpp:106] Creating Layer loss_cls
I0301 13:14:28.733455 26256 net.cpp:454] loss_cls <- cls_score
I0301 13:14:28.733462 26256 net.cpp:454] loss_cls <- labels
I0301 13:14:28.733475 26256 net.cpp:411] loss_cls -> loss_cls
I0301 13:14:28.733489 26256 layer_factory.hpp:77] Creating layer loss_cls
I0301 13:14:28.734452 26256 net.cpp:150] Setting up loss_cls
I0301 13:14:28.734465 26256 net.cpp:157] Top shape: (1)
I0301 13:14:28.734470 26256 net.cpp:160]     with loss weight 1
I0301 13:14:28.734501 26256 net.cpp:165] Memory required for data: 501106520
I0301 13:14:28.734508 26256 layer_factory.hpp:77] Creating layer loss_bbox
I0301 13:14:28.734521 26256 net.cpp:106] Creating Layer loss_bbox
I0301 13:14:28.734527 26256 net.cpp:454] loss_bbox <- bbox_pred
I0301 13:14:28.734534 26256 net.cpp:454] loss_bbox <- bbox_targets
I0301 13:14:28.734541 26256 net.cpp:454] loss_bbox <- bbox_inside_weights
I0301 13:14:28.734547 26256 net.cpp:454] loss_bbox <- bbox_outside_weights
I0301 13:14:28.734556 26256 net.cpp:411] loss_bbox -> loss_bbox
I0301 13:14:28.734704 26256 net.cpp:150] Setting up loss_bbox
I0301 13:14:28.734712 26256 net.cpp:157] Top shape: (1)
I0301 13:14:28.734716 26256 net.cpp:160]     with loss weight 1
I0301 13:14:28.734724 26256 net.cpp:165] Memory required for data: 501106524
I0301 13:14:28.734730 26256 net.cpp:226] loss_bbox needs backward computation.
I0301 13:14:28.734737 26256 net.cpp:226] loss_cls needs backward computation.
I0301 13:14:28.734742 26256 net.cpp:226] bbox_pred needs backward computation.
I0301 13:14:28.734747 26256 net.cpp:226] cls_score needs backward computation.
I0301 13:14:28.734752 26256 net.cpp:226] fc7_drop7_0_split needs backward computation.
I0301 13:14:28.734757 26256 net.cpp:226] drop7 needs backward computation.
I0301 13:14:28.734762 26256 net.cpp:226] relu7 needs backward computation.
I0301 13:14:28.734766 26256 net.cpp:226] fc7 needs backward computation.
I0301 13:14:28.734771 26256 net.cpp:226] drop6 needs backward computation.
I0301 13:14:28.734777 26256 net.cpp:226] relu6 needs backward computation.
I0301 13:14:28.734781 26256 net.cpp:226] fc6 needs backward computation.
I0301 13:14:28.734786 26256 net.cpp:226] roi_pool5 needs backward computation.
I0301 13:14:28.734792 26256 net.cpp:226] roi-data needs backward computation.
I0301 13:14:28.734802 26256 net.cpp:226] proposal needs backward computation.
I0301 13:14:28.734813 26256 net.cpp:226] rpn_cls_prob_reshape needs backward computation.
I0301 13:14:28.734819 26256 net.cpp:226] rpn_cls_prob needs backward computation.
I0301 13:14:28.734825 26256 net.cpp:226] rpn_loss_bbox needs backward computation.
I0301 13:14:28.734833 26256 net.cpp:226] rpn_loss_cls needs backward computation.
I0301 13:14:28.734840 26256 net.cpp:226] rpn-data needs backward computation.
I0301 13:14:28.734853 26256 net.cpp:226] rpn_cls_score_reshape_rpn_cls_score_reshape_0_split needs backward                              computation.
I0301 13:14:28.734860 26256 net.cpp:226] rpn_cls_score_reshape needs backward computation.
I0301 13:14:28.734866 26256 net.cpp:226] rpn_bbox_pred_rpn_bbox_pred_0_split needs backward computation.
I0301 13:14:28.734872 26256 net.cpp:226] rpn_bbox_pred needs backward computation.
I0301 13:14:28.734877 26256 net.cpp:226] rpn_cls_score_rpn_cls_score_0_split needs backward computation.
I0301 13:14:28.734886 26256 net.cpp:226] rpn_cls_score needs backward computation.
I0301 13:14:28.734894 26256 net.cpp:226] rpn/output_rpn_relu/3x3_0_split needs backward computation.
I0301 13:14:28.734900 26256 net.cpp:226] rpn_relu/3x3 needs backward computation.
I0301 13:14:28.734905 26256 net.cpp:226] rpn_conv/3x3 needs backward computation.
I0301 13:14:28.734911 26256 net.cpp:226] conv5_relu5_0_split needs backward computation.
I0301 13:14:28.734917 26256 net.cpp:226] relu5 needs backward computation.
I0301 13:14:28.734921 26256 net.cpp:226] conv5 needs backward computation.
I0301 13:14:28.734927 26256 net.cpp:226] relu4 needs backward computation.
I0301 13:14:28.734931 26256 net.cpp:226] conv4 needs backward computation.
I0301 13:14:28.734937 26256 net.cpp:226] relu3 needs backward computation.
I0301 13:14:28.734942 26256 net.cpp:226] conv3 needs backward computation.
I0301 13:14:28.734946 26256 net.cpp:226] pool2 needs backward computation.
I0301 13:14:28.734951 26256 net.cpp:226] norm2 needs backward computation.
I0301 13:14:28.734957 26256 net.cpp:226] relu2 needs backward computation.
I0301 13:14:28.734962 26256 net.cpp:226] conv2 needs backward computation.
I0301 13:14:28.734968 26256 net.cpp:228] pool1 does not need backward computation.
I0301 13:14:28.734977 26256 net.cpp:228] norm1 does not need backward computation.
I0301 13:14:28.734987 26256 net.cpp:228] relu1 does not need backward computation.
I0301 13:14:28.734992 26256 net.cpp:228] conv1 does not need backward computation.
I0301 13:14:28.734997 26256 net.cpp:228] gt_boxes_input-data_2_split does not need backward computation.
I0301 13:14:28.735004 26256 net.cpp:228] im_info_input-data_1_split does not need backward computation.
I0301 13:14:28.735011 26256 net.cpp:228] data_input-data_0_split does not need backward computation.
I0301 13:14:28.735018 26256 net.cpp:228] input-data does not need backward computation.
I0301 13:14:28.735023 26256 net.cpp:270] This network produces output loss_bbox
I0301 13:14:28.735028 26256 net.cpp:270] This network produces output loss_cls
I0301 13:14:28.735034 26256 net.cpp:270] This network produces output rpn_cls_loss
I0301 13:14:28.735041 26256 net.cpp:270] This network produces output rpn_loss_bbox
I0301 13:14:28.735111 26256 net.cpp:283] Network initialization done.
I0301 13:14:28.735357 26256 solver.cpp:60] Solver scaffolding done.
Loading pretrained model weights from data/imagenet_models/VGG_CNN_M_1024_Scale.v2.caffemodel
I0301 13:14:30.426005 26256 net.cpp:816] Ignoring source layer pool5
I0301 13:14:30.504096 26256 net.cpp:816] Ignoring source layer fc8
I0301 13:14:30.504124 26256 net.cpp:816] Ignoring source layer prob
Solving...
I0301 13:14:30.943373 26256 solver.cpp:229] Iteration 0, loss = 1.71672
I0301 13:14:30.943447 26256 solver.cpp:245]     Train net output #0: loss_bbox = 1.40663e-05 (* 1 = 1.40663                             e-05 loss)
I0301 13:14:30.943457 26256 solver.cpp:245]     Train net output #1: loss_cls = 1.00689 (* 1 = 1.00689 loss                             )
I0301 13:14:30.943464 26256 solver.cpp:245]     Train net output #2: rpn_cls_loss = 0.699558 (* 1 = 0.69955                             8 loss)
I0301 13:14:30.943470 26256 solver.cpp:245]     Train net output #3: rpn_loss_bbox = 0.0102539 (* 1 = 0.010                             2539 loss)
I0301 13:14:30.943478 26256 sgd_solver.cpp:106] Iteration 0, lr = 0.001
I0301 13:14:35.436295 26256 solver.cpp:229] Iteration 20, loss = 0.469994
I0301 13:14:35.436348 26256 solver.cpp:245]     Train net output #0: loss_bbox = 0.00194004 (* 1 = 0.001940                             04 loss)
I0301 13:14:35.436357 26256 solver.cpp:245]     Train net output #1: loss_cls = 0.285266 (* 1 = 0.285266 lo                             ss)
I0301 13:14:35.436362 26256 solver.cpp:245]     Train net output #2: rpn_cls_loss = 0.148352 (* 1 = 0.14835                             2 loss)
I0301 13:14:35.436368 26256 solver.cpp:245]     Train net output #3: rpn_loss_bbox = 0.0344359 (* 1 = 0.034                             4359 loss)
I0301 13:14:35.436376 26256 sgd_solver.cpp:106] Iteration 20, lr = 0.001
I0301 13:14:39.824944 26256 solver.cpp:229] Iteration 40, loss = 0.284742
I0301 13:14:39.825048 26256 solver.cpp:245]     Train net output #0: loss_bbox = 0.136684 (* 1 = 0.136684 l                             oss)
I0301 13:14:39.825071 26256 solver.cpp:245]     Train net output #1: loss_cls = 0.12873 (* 1 = 0.12873 loss                             )
I0301 13:14:39.825100 26256 solver.cpp:245]     Train net output #2: rpn_cls_loss = 0.0182598 (* 1 = 0.0182                             598 loss)
I0301 13:14:39.825115 26256 solver.cpp:245]     Train net output #3: rpn_loss_bbox = 0.00106758 (* 1 = 0.00                             106758 loss)
_LYH_
关注
  • 0
    点赞
  • 踩
  • 0
    收藏
    觉得还不错? 一键收藏
  • 0
    评论
专栏目录
论文笔记 / Mitosis Detection in Breast Cancer Histology Images with Deep Neural Networks
weixin_43404454的博客
12-13 1102
仅供参考，如有翻译不到位的地方敬请指出。 论文地址：https://link.springer.com/chapter/10.1007/978-3-642-40763-5_51 摘要 我们使用含有最大池化层的深度卷积神经网络来检测乳腺组织学图像中的有丝分裂。训练网络以使用以像素为中心的patch作为上下文对图像中的每个像素进行分类。 然后将简单的后处理应用于网络的输出。我们的方法赢得了ICPR 2...
Cell mitosis detection using deep neural networks
02-07
Automatic cell mitosis detection can greatly facilitate the investigation of cell life cycle. However, cell-type diversity, cell non-rigid deformation and high cell density pose difficulties on ...
mitosis detection
qq_41574648的博客
02-27 452
问题一 ./experiments/scripts/faster_rcnn_end2end_12.sh 0 VGG_CNN_M_1024_Scale mitos --set EXP_DIR XX + set -e + export PYTHONUNBUFFERED=True + PYTHONUNBUFFERED=True + GPU_ID=0 + NET=VGG_CNN_M_1024_Scale...
DeepMitosis: Mitosis detection via deep detection, verification and segmentation networks 论文解读
qq_41574648的博客
02-27 363
3. Methods 3.1. Using deep detection network to detect mitosis 1） anchor采用默认 The anchors of RPN have three scales and three aspect ratios. In this application, the shape of mitosis is irregular, maki...
mitosis detection（二）
qq_41574648的博客
03-01 304
I0301 12:36:30.536449 25955 solver.cpp:245] Train net output #0: loss_bbox = 0.00237033 (* 1 = 0.00237033 loss) I0301 12:36:30.536458 25955 solver.cpp:245] Train net output #1: loss_cls = 0.06...
A Unified Framework for Tumor ProliferationScore Prediction in Breast Histopathology
卓为的博客
01-31 330
A Unified Framework for Tumor ProliferationScore Prediction in Breast Histopathology This paper was accepted by the 3rd Workshop on Deep Learning in Medical Image Analysis (DLMIA 2017), MICCAI 2017.
Spatiotemporal Joint Mitosis Detection Using CNN-LSTM Network in Time-Lapse Phase Contrast Microscopy Images
02-07
### Spatiotemporal Joint Mitosis Detection Using CNN-LSTM Network #### 概述 本文介绍了一种用于在时间延时相位对比显微镜图像中同时进行空间和时间上细胞有丝分裂检测的方法。该方法结合了卷积神经网络...
图像分割实验代码matlab-SegMitos_mitosis_detection:使用同心损失在乳腺组织病理学图像中的弱监督有丝分裂检测
05-27
SegMitos_mitosis_detection 使用同心损失弱监督的乳腺病理组织学图像中的有丝分裂检测。 我们的医学图像分析论文的编码。 内容 要求：软件 我们使用深度学习框架Caffe来训练细分网络。 用于数据准备的代码是由...
Mitosis-开源
04-27
用C ++编写的可自定义的蜂窝自动机程序
Mitosis Detection in Breast Cancer Histology Images with Deep Neural Networks
卓为的博客
12-30 1086
1，Mitosis Detection in Breast Cancer Histology Images with Deep Neural Networks 这篇文章发表在 MICCAI2013 ,文章中提出的方法在 ICPR2012 mitosis detection competition 中取得了第一名。 其中方法大致为(Fig.1)： Fig.1 1,训练一个CNN分类器
数据竞赛综述
Solomon1558的专栏
07-29 1802
数据竞赛综述    随着大数据时代的到来，海量的数据驱动着机器学习的高速发展，构建相应的测试评估体系势在必行。数据建模和分析挑战赛主要由以下几个要素构成： * 数据集：一般包括训练集、验证集和测试集。其中训练集和验证集是公开的，包含原始图像和标注信息（物体的边框坐标、类别标签），分别用于模型训练和模型选择。测试集不对外公布，用于评估算法的精度。 * 开发者工具/文档（可选）：辅助参赛者使用数据集
2019.7学习总结-目标检测-Python+pytorch
qq_33666011的博客
07-01 984
目录 1、基础知识 1.1、dict的copy: 1.2、torch.cuda.synchronize() 1.3、在CenterNet里 1.4、torch.gather、topk用法： 1.5、torch常用的函数 1.6 torch.gather实现找3维度的Index 2、kernel 2.1、核函数 2.2 不同的核函数 1、基础知识 1.1、dict的c...
计算机视觉（三）
程序猿的视界
12-26 1125
1、滤波器 傅里叶变换主要作用反应图像各区域像素变化的幅度。 滤波器（核）矩阵：一组权重集合（内部所有值加和为0），作用在源图像的一个区域（滑动），并由此生成目标图像的一个元素。 高通滤波器（HPF）：检测图像的某个区域，然后根据像素与周围像素的亮度差值来提升该像素的滤波器。如果亮度变化很大，中央像素的亮度会增加（反之则不会）。 （即某个像素比它周围的像素更突出，就会提升它的亮度）
数字图像处理笔记（二）：使用OpenCV处理图像
HHH_ANS的博客
12-06 1010
1 - 傅里叶变换
吴恩达的机器学习编程作业12：gaussianKernel 高斯核函数
melon__的博客
08-28 5503
function sim = gaussianKernel(x1, x2, sigma) %RBFKERNEL returns a radial basis function kernel between x1 and x2 % sim = gaussianKernel(x1, x2) returns a gaussian kernel between x1 and x2 % and r...
利用Caffe做回归(regression)
weixin_30896763的博客
03-30 448
2017-02-22，Update： 鉴于这篇文章反馈很多，有部分读者照着这篇文章步骤的读者跑不通来咨询，因为这篇里代码是分散写的很容易漏掉步骤，如果完全按照文章是可以重现的，如果还是不能重现，有个简化后的例子，全部代码在这里： https://github.com/frombeijingwithlove/dlcv_for_beginners/tree/master/chap9 按照README....
Geotrellis学习笔记（一）——核密度分析
热门推荐
qq_32432081的博客
06-27 2万+
在geotrellis环境下成功运行了helloworld之后，我第一个尝试的核密度计算~整个过程还是挺艰难的。。。因为对scala非常地不熟，基本属于边写边学的状态T^T 嗯。。首先 核密度分析是什么？？？ 官方文档里对核密度分析有一段这样的介绍：        Kernel density is one way to convert a set of points (an instanc...
OpenCV实现二维高斯核GaussianKernel
Emily的博客
09-14 1万+
matlab： matlab示例 h = fspecial(‘gaussian’, hsize, sigma) returns a rotationally symmetric Gaussian lowpass filter of size hsize with standard deviation sigma (positive). hsize can be a vector spe...
评论
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。

余额充值