VGG16 finetuned on the Salient Object Subitizing (SOS) dataset, which is described in the CVPR'15 paper: "Salient Object Subitizing"

README.md

##Information

name: VGG16 finetuned on the Salient Object Subitizing (SOS) dataset, which is described in the CVPR'15 paper: "Salient Object Subitizing"

caffemodel: VGG16_SalObjSub.caffemodel

caffemodel_url: http://www.cs.bu.edu/groups/ivc/data/SOS/VGG16_SalObjSub.caffemodel

license: http://creativecommons.org/licenses/by-nc/4.0/ (non-commercial use only)

caffe_version: trained using a custom Caffe-based framework (see https://gist.github.com/ksimonyan/211839e770f7b538e2d8#file-readme-md)

gist_id: 27c1c0a7736ba66c2395

Description

The model predicts the existence and the number (0, 1, 2, 3 and 4+) of salient objects in an image:

Salient Object Subitizing
J. Zhang, S. Ma, M. Sameki, S. Sclaroff, M. Betke, Z. Lin, X. Shen, B. Price and R. Mech. 
CVPR, 2015.

Please cite the paper if you use the model. See our project page for more details:

http://www.cs.bu.edu/groups/ivc/Subitizing/

The input images should be zero-centered by mean pixel (rather than mean image) subtraction. Namely, the following BGR values should be subtracted: [103.939, 116.779, 123.68].

Caffe compatibility

see https://gist.github.com/ksimonyan/211839e770f7b538e2d8#file-readme-md

How to use the model

The model outputs a 5D probability vector for a test image. The each value in the 5D vector corresponds to the likelihood that this image contains 0, 1, 2, 3 and 4+ salient objects. In our experiments, test images are resized to 224X224, regardless of the original aspect ratios.

For example applications in salient object detection and object proposal generation, please see the paper.

deploy.prototxt

name: "VGG_ILSVRC_16_layers"
input: "data"
input_dim: 10
input_dim: 3
input_dim: 224
input_dim: 224
layers {
  bottom: "data"
  top: "conv1_1"
  name: "conv1_1"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 64
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv1_1"
  top: "conv1_1"
  name: "relu1_1"
  type: RELU
}
layers {
  bottom: "conv1_1"
  top: "conv1_2"
  name: "conv1_2"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 64
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv1_2"
  top: "conv1_2"
  name: "relu1_2"
  type: RELU
}
layers {
  bottom: "conv1_2"
  top: "pool1"
  name: "pool1"
  type: POOLING
  pooling_param {
    pool: MAX
    kernel_size: 2
    stride: 2
  }
}
layers {
  bottom: "pool1"
  top: "conv2_1"
  name: "conv2_1"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 128
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv2_1"
  top: "conv2_1"
  name: "relu2_1"
  type: RELU
}
layers {
  bottom: "conv2_1"
  top: "conv2_2"
  name: "conv2_2"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 128
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv2_2"
  top: "conv2_2"
  name: "relu2_2"
  type: RELU
}
layers {
  bottom: "conv2_2"
  top: "pool2"
  name: "pool2"
  type: POOLING
  pooling_param {
    pool: MAX
    kernel_size: 2
    stride: 2
  }
}
layers {
  bottom: "pool2"
  top: "conv3_1"
  name: "conv3_1"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 256
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv3_1"
  top: "conv3_1"
  name: "relu3_1"
  type: RELU
}
layers {
  bottom: "conv3_1"
  top: "conv3_2"
  name: "conv3_2"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 256
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv3_2"
  top: "conv3_2"
  name: "relu3_2"
  type: RELU
}
layers {
  bottom: "conv3_2"
  top: "conv3_3"
  name: "conv3_3"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 256
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv3_3"
  top: "conv3_3"
  name: "relu3_3"
  type: RELU
}
layers {
  bottom: "conv3_3"
  top: "pool3"
  name: "pool3"
  type: POOLING
  pooling_param {
    pool: MAX
    kernel_size: 2
    stride: 2
  }
}
layers {
  bottom: "pool3"
  top: "conv4_1"
  name: "conv4_1"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 512
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv4_1"
  top: "conv4_1"
  name: "relu4_1"
  type: RELU
}
layers {
  bottom: "conv4_1"
  top: "conv4_2"
  name: "conv4_2"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 512
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv4_2"
  top: "conv4_2"
  name: "relu4_2"
  type: RELU
}
layers {
  bottom: "conv4_2"
  top: "conv4_3"
  name: "conv4_3"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 512
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv4_3"
  top: "conv4_3"
  name: "relu4_3"
  type: RELU
}
layers {
  bottom: "conv4_3"
  top: "pool4"
  name: "pool4"
  type: POOLING
  pooling_param {
    pool: MAX
    kernel_size: 2
    stride: 2
  }
}
layers {
  bottom: "pool4"
  top: "conv5_1"
  name: "conv5_1"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 512
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv5_1"
  top: "conv5_1"
  name: "relu5_1"
  type: RELU
}
layers {
  bottom: "conv5_1"
  top: "conv5_2"
  name: "conv5_2"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 512
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv5_2"
  top: "conv5_2"
  name: "relu5_2"
  type: RELU
}
layers {
  bottom: "conv5_2"
  top: "conv5_3"
  name: "conv5_3"
  type: CONVOLUTION
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  convolution_param {
    num_output: 512
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv5_3"
  top: "conv5_3"
  name: "relu5_3"
  type: RELU
}
layers {
  bottom: "conv5_3"
  top: "pool5"
  name: "pool5"
  type: POOLING
  pooling_param {
    pool: MAX
    kernel_size: 2
    stride: 2
  }
}
layers {
  bottom: "pool5"
  top: "fc6"
  name: "fc6"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: INNER_PRODUCT
  inner_product_param {
    num_output: 4096
  }
}
layers {
  bottom: "fc6"
  top: "fc6"
  name: "relu6"
  type: RELU
}
layers {
  bottom: "fc6"
  top: "fc6"
  name: "drop6"
  type: DROPOUT
  dropout_param {
    dropout_ratio: 0.5
  }
}
layers {
  bottom: "fc6"
  top: "fc7"
  name: "fc7"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: INNER_PRODUCT
  inner_product_param {
    num_output: 4096
  }
}
layers {
  bottom: "fc7"
  top: "fc7"
  name: "relu7"
  type: RELU
}
layers {
  bottom: "fc7"
  top: "fc7"
  name: "drop7"
  type: DROPOUT
  dropout_param {
    dropout_ratio: 0.5
  }
}
layers {
  bottom: "fc7"
  top: "fc8"
  name: "fc8-5"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: INNER_PRODUCT
  inner_product_param {
    num_output: 5
  }
}
layers {
  bottom: "fc8"
  top: "prob"
  name: "prob"
  type: SOFTMAX
}

solver.prototxt

net: "train_val.prototxt"
test_iter: 120
test_interval: 2000
base_lr: 0.0001
lr_policy: "step"
gamma: 0.1
stepsize: 5000
display: 50
max_iter: 450000
momentum: 0.9
weight_decay: 0.0005
snapshot: 10000
snapshot_prefix: "snapshot_full/VGG16_imagenet_train"
solver_mode: GPU

train_val.prototxt

name: "VGG_ILSVRC_16_layers"
layers {
  top: "data"
  top: "label"
  name: "data"
  type: DATA
  data_param {
    source: "/research/cbi/jmzhang_work_dir/caffe-openhero/examples/imagenet_sbt_finetune/sbt_train_leveldb"
    backend: LEVELDB
    batch_size: 8
  }
  image_data_param {
    shuffle: true
    new_height: 256
    new_width: 256
  }
  transform_param {
    crop_size: 224
    mirror: true
    mean_file: "../../data/ilsvrc12/imagenet_mean.binaryproto"
  }
  include: { phase: TRAIN }
}
layers {
  top: "data"
  top: "label"
  name: "data"
  type: DATA
  data_param {
    source: "/research/cbi/jmzhang_work_dir/caffe-openhero/examples/imagenet_sbt_finetune/sbt_val_leveldb"
    backend: LEVELDB
    batch_size: 10
  }
  image_data_param {
    new_height: 256
    new_width: 256
  }
  transform_param {
    crop_size: 224
    mirror: false
    mean_file: "../../data/ilsvrc12/imagenet_mean.binaryproto"
  }
  include: { phase: TEST }
}
layers {
  bottom: "data"
  top: "conv1_1"
  name: "conv1_1"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 64
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv1_1"
  top: "conv1_1"
  name: "relu1_1"
  type: RELU
}
layers {
  bottom: "conv1_1"
  top: "conv1_2"
  name: "conv1_2"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 64
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv1_2"
  top: "conv1_2"
  name: "relu1_2"
  type: RELU
}
layers {
  bottom: "conv1_2"
  top: "pool1"
  name: "pool1"
  type: POOLING
  pooling_param {
    pool: MAX
    kernel_size: 2
    stride: 2
  }
}
layers {
  bottom: "pool1"
  top: "conv2_1"
  name: "conv2_1"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 128
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv2_1"
  top: "conv2_1"
  name: "relu2_1"
  type: RELU
}
layers {
  bottom: "conv2_1"
  top: "conv2_2"
  name: "conv2_2"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 128
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv2_2"
  top: "conv2_2"
  name: "relu2_2"
  type: RELU
}
layers {
  bottom: "conv2_2"
  top: "pool2"
  name: "pool2"
  type: POOLING
  pooling_param {
    pool: MAX
    kernel_size: 2
    stride: 2
  }
}
layers {
  bottom: "pool2"
  top: "conv3_1"
  name: "conv3_1"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 256
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv3_1"
  top: "conv3_1"
  name: "relu3_1"
  type: RELU
}
layers {
  bottom: "conv3_1"
  top: "conv3_2"
  name: "conv3_2"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 256
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv3_2"
  top: "conv3_2"
  name: "relu3_2"
  type: RELU
}
layers {
  bottom: "conv3_2"
  top: "conv3_3"
  name: "conv3_3"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 256
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv3_3"
  top: "conv3_3"
  name: "relu3_3"
  type: RELU
}
layers {
  bottom: "conv3_3"
  top: "pool3"
  name: "pool3"
  type: POOLING
  pooling_param {
    pool: MAX
    kernel_size: 2
    stride: 2
  }
}
layers {
  bottom: "pool3"
  top: "conv4_1"
  name: "conv4_1"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 512
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv4_1"
  top: "conv4_1"
  name: "relu4_1"
  type: RELU
}
layers {
  bottom: "conv4_1"
  top: "conv4_2"
  name: "conv4_2"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 512
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv4_2"
  top: "conv4_2"
  name: "relu4_2"
  type: RELU
}
layers {
  bottom: "conv4_2"
  top: "conv4_3"
  name: "conv4_3"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 512
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv4_3"
  top: "conv4_3"
  name: "relu4_3"
  type: RELU
}
layers {
  bottom: "conv4_3"
  top: "pool4"
  name: "pool4"
  type: POOLING
  pooling_param {
    pool: MAX
    kernel_size: 2
    stride: 2
  }
}
layers {
  bottom: "pool4"
  top: "conv5_1"
  name: "conv5_1"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 512
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv5_1"
  top: "conv5_1"
  name: "relu5_1"
  type: RELU
}
layers {
  bottom: "conv5_1"
  top: "conv5_2"
  name: "conv5_2"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: CONVOLUTION
  convolution_param {
    num_output: 512
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv5_2"
  top: "conv5_2"
  name: "relu5_2"
  type: RELU
}
layers {
  bottom: "conv5_2"
  top: "conv5_3"
  name: "conv5_3"
  type: CONVOLUTION
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  convolution_param {
    num_output: 512
    pad: 1
    kernel_size: 3
  }
}
layers {
  bottom: "conv5_3"
  top: "conv5_3"
  name: "relu5_3"
  type: RELU
}
layers {
  bottom: "conv5_3"
  top: "pool5"
  name: "pool5"
  type: POOLING
  pooling_param {
    pool: MAX
    kernel_size: 2
    stride: 2
  }
}
layers {
  bottom: "pool5"
  top: "fc6"
  name: "fc6"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: INNER_PRODUCT
  inner_product_param {
    num_output: 4096
  }
}
layers {
  bottom: "fc6"
  top: "fc6"
  name: "relu6"
  type: RELU
}
layers {
  bottom: "fc6"
  top: "fc6"
  name: "drop6"
  type: DROPOUT
  dropout_param {
    dropout_ratio: 0.5
  }
}
layers {
  bottom: "fc6"
  top: "fc7"
  name: "fc7"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: INNER_PRODUCT
  inner_product_param {
    num_output: 4096
  }
}
layers {
  bottom: "fc7"
  top: "fc7"
  name: "relu7"
  type: RELU
}
layers {
  bottom: "fc7"
  top: "fc7"
  name: "drop7"
  type: DROPOUT
  dropout_param {
    dropout_ratio: 0.5
  }
}
layers {
  bottom: "fc7"
  top: "fc8"
  name: "fc8-5"
  blobs_lr: 1
  blobs_lr: 2
  weight_decay: 1
  weight_decay: 0
  type: INNER_PRODUCT
  inner_product_param {
    num_output: 5
  }
}
layers {
  name: "accuracy"
  type: ACCURACY
  bottom: "fc8"
  bottom: "label"
  top: "accuracy"
  include: { phase: TEST }
}
layers {
  bottom: "fc8"
  bottom: "label"
  name: "loss"
  type: SOFTMAX_LOSS
  include: { phase: TRAIN }
}