Keras model classifying new images in a folder on docker
Edit dockerfile to provide paths for motionEye volume.
The serving
and serving_no_avx
folder contain dockerfiles for Tensorflow Serving
containers used by the application.
Most CPUs support the AVX instruction set, but if your host runs on a Celeron, or you use a hypervisor that does not provide AVX you'd need the no_avx
backend.
The docker_compose.yml
files are meant to be run with Docker Swarm. If the Visualizer is not needed, remove the respective part of the config. Also, edit volume mount paths, to where the model files will be stored on nodes.