Tensorflow Object Detection API for Windows
Tensorflow recently released a new object detection api Is there a way to run this on windows? Directions for Linux.
source to share
We do not officially support the Tensorflow Object Discovery API, but some external users got it working.
Our dependencies are pillow, lxml, jupyter, matplotlib and protobuf. You can download the protobuf compiler version here . The rest of the dependencies can be installed using pip.
source to share
As I said in another post , you can use your local GPU on Windows as Tensorflow supports GPUs in python.
And here is an example.
Unfortunately Tensoflow does not support tensorflow-serving
in windows. Also, as you said, is Nvidia-Docker
not supported on windows. Bash on windows
does not support GPU. So I think this is the only easy way right now.
source to share
Yes, you can run the Tensorflow Object Discovery API on Windows. Unfortunately, this is a bit tricky and the official documentation does not reflect it properly. I used the following procedure:
- Install Tensorflow natively on Windows using Anaconda + CUDA + cuDNN . Note that TF 1.5 is now built against CUDA 9.0, so be sure to download the appropriate versions.
-
Then you clone the repository and create the Protobuf files as described in the tutorial, but be careful, there is a bug in Windows Protobuf 3.5 , so make sure you are using version 3.4 .
cd [TF-models]\research protoc.exe object_detection/protos/*.proto --python_out=.
-
Finally, you need to build and install packages with
cd [TF-models]\research\slim python setup.py install cd [TF-models]\research python setup.py install
If you get an exception error: failed to create "BUILD": unable to create file, if this file already exists, first delete the BUILD file, it will be automatically created
-
And make the inline binaries available to your python path or just copy the directories
slim
andobject_detection
into the directory[Anaconda3]/Lib/site-packages
To see everything collected, check out our music object detector , which was trained in Windows and Linux.
source to share