SeaOfOcean

name: "PVANET"

################################################################################
## Input
################################################################################

input: "data"
input_shape {
  dim: 1
  dim: 3

SeaOfOcean

name: "PVANET finetune"
layer {
  name: 'input-data'
  type: 'Python'
  top: 'data'
  top: 'im_info'
  top: 'gt_boxes'
  python_param {
    module: 'roi_data_layer.layer'
    layer: 'RoIDataLayer'

SeaOfOcean

name: "PVANet train"
layer {
  name: 'input-data'
  type: 'Python'
  top: 'data'
  top: 'im_info'
  top: 'gt_boxes'
  python_param {
    module: 'roi_data_layer.layer'
    layer: 'RoIDataLayer'

SeaOfOcean

name: "PVANET test"

################################################################################
## Input
################################################################################

input: "data"
input_shape {
  dim: 1
  dim: 3

SeaOfOcean

name: "VGG_ILSVRC_16_layers"
layer {
  name: 'input-data'
  type: 'Python'
  top: 'data'
  top: 'im_info'
  top: 'gt_boxes'
  python_param {
    module: 'roi_data_layer.layer'
    layer: 'RoIDataLayer'

SeaOfOcean

name: "VGG_ILSVRC_16_layers"
layer {
  name: 'input-data'
  type: 'Python'
  top: 'data'
  top: 'im_info'
  top: 'gt_boxes'
  python_param {
    module: 'roi_data_layer.layer'
    layer: 'RoIDataLayer'

SeaOfOcean

1. clone analytics-zoo project
```bash
git clone https://github.com/intel-analytics/analytics-zoo.git
```

2. install image transformer library
```
# mvn install image transformer lib
cd transform/vision
mvn clean install

SeaOfOcean

from bigdl.util.common import *
init_engine()

SeaOfOcean

import cv2

folder = "images"
imageFiles = os.listdir(folder)
images = []
for f in imageFiles:
    image = cv2.imread(folder + '/' + f)
    images.append(image)

img_rdd = sc.parallelize(images)

SeaOfOcean

from bigdl.nn.layer import Model
# load model
model = Model.load("bigdl_mobilenet.model")
print 'load model done'