Cuda-Chen / Mask_RCNN

Mask R-CNN for object detection and instance segmentation on Keras and TensorFlow

Mask RCNN

Mask R-CNN for fish image segmentation.

How to Set Up

via conda:

$ conda create -n maskrcnn python=3.6 pip
$ conda activate maskrcnn
$ pip install -r requirements.txt
$ python setup.py install

via Docker:
If you are too lazy to manage the virtual environment, you can pull this Docker image built by me (CPU-only):

$ docker run -it -p 8888:8888 -p 6006:6006 -v ~/:/host cudachen/mask-rcnn-docker

Note the -v option. It maps your user directory (~/) to /host in the container. Change it if needed. The two -p options expose the ports used by Jupyter Notebook and Tensorboard, respectively.

Or setting up Docker with Jupyter Notebook (be cautious of security issues, see this):

$ docker run -it -p 8888:8888 -p 6006:6006 -v ~/:/host cudachen/mask-rcnn-docker jupyter notebook --allow-root /host

then navigate to http://localhost:8888/.