examples/object_detection
Daniel Castellanos e6b6730650 Updated object detection training example (#228)
* Updated Dockerfile.traning to use latest tensorflow
  and tensorflow object detetion api.
* Updated tf-training-job component and added a chief
  replica spec
* Corrected some typos and updated some instructions
2018-08-20 19:32:12 -07:00
..
batch-predict Batch predict example for object detection using GPU (#199) 2018-08-03 11:57:53 -07:00
conf Updated object detection training example (#228) 2018-08-20 19:32:12 -07:00
docker Updated object detection training example (#228) 2018-08-20 19:32:12 -07:00
ks-app Updated object detection training example (#228) 2018-08-20 19:32:12 -07:00
serving_script Example of TF Serving with GPU (#154) 2018-07-24 21:44:55 -07:00
README.md Batch predict example for object detection using GPU (#199) 2018-08-03 11:57:53 -07:00
export_tf_graph.md Updated object detection training example (#228) 2018-08-20 19:32:12 -07:00
monitor_job.md Updated object detection training example (#228) 2018-08-20 19:32:12 -07:00
setup.md Updated object detection training example (#228) 2018-08-20 19:32:12 -07:00
submit_batch_predict.md Batch predict example for object detection using GPU (#199) 2018-08-03 11:57:53 -07:00
submit_job.md Fixed broken links in object detection example (#211) 2018-08-03 16:05:27 -07:00
tf_serving_cpu.md Fixed broken links in object detection example (#211) 2018-08-03 16:05:27 -07:00
tf_serving_gpu.md Small fix to TF serving gpu (#221) 2018-08-14 14:27:35 -07:00

README.md

Distributed TensorFlow Object Detection Training and Serving on K8s with Kubeflow

This example demonstrates how to use kubeflow to train/serve an object detection model on an existing K8s cluster using the TensorFlow object detection API

This example is based on the TensorFlow Pets tutorial.

Steps:

  1. Setup a Kubeflow cluster
  2. Submit a distributed object detection training job
  3. Monitor your training job
  4. Export model
  5. Serve the model with GPU
  6. Submit a batch prediction job using GPU