參考：

概述如何安裝

Step 1: create a conda virtual environment with python 3.6
Step 2: install the dependencies
Step 3: Clone the Mask_RCNN repo
Step 4: install pycocotools
Step 5: download the pre-trained weights
Step 6: Test it

Step 1 - Create a conda virtual environment

we will be using Anaconda with python 3.6.
If you don’t have Anaconda, follow this tutorial

• run this command in a CMD window

conda create -n MaskRCNN python=3.6 pip

Step 2 - Install the Dependencies

actvitate MaskRCNN
pip install -r requirements.txt

• NOTE: we’re installing these (tf-gpu requires some pre-reqs)
  numpy, scipy, cython, h5py, Pillow, scikit-image,
  tensorflow-gpu==1.5, keras, jupyter

Step 3 - Clone the Mask RCNN Repo

• Run this command

git clone https://github.com/matterport/Mask_RCNN.git

Step 4 - Install pycocotools

git clone https://github.com/philferriere/cocoapi.git

• use pip to install pycocotools

pip install git+https://github.com/philferriere/cocoapi.git#subdirectory=PythonAPI
 

Step 5 - Download the Pre-trained Weights

Step 6 - Let’s Test it!

open up the demo.ipynb and run it