[DOCKER] I tried moving EfficientDet
What do you want to do
--I want to move EfficientDet ――It takes a lot of time to do Evalation at COCO, so even infer one image.
Caution
-Repository to use belongs to google, but it is not an official google product (reference automl / tree / master / coefficientdet # 6-training-efficientdets-on-tpus)) --Since the code is being updated rapidly, this information will soon be outdated as well.
environment
- Docker 18.09.1
- Ubuntu 16.04
- GTX1080
It worked when I did this
For the time being, from the console
docker pull tensorflow/tensorflow:latest-devel-gpu-py3
docker run -it --name tensorflow -p 8888:8888 tensorflow/tensorflow:latest-devel-gpu-py3 bash
pip install jupyter
git clone https://github.com/google/automl
cd automl
jupyter notebook --port 8888 --ip=0.0.0.0 --allow-root
# 127.0.0.Open ~ in your browser
Tutorial from the browser.open ipynb
#pycocotools does not work with python3 series, so fix it a little
# !pip install -U 'git+https://github.com/cocodataset/cocoapi.git#subdirectory=PythonAPI'
#Overwrite the part with ↓
!git clone https://github.com/cocodataset/cocoapi
!cd cocoapi/PythonAPI && make
!pip install Cython
!pip install pycocotools
After that, hit Shift + Enter and wait for this image.
Recommended Posts