Modeling Guide

Infer MNIST Stream UI

This graph applies a neuronal network based on the tensorflow MNIST example below on a stream of images of handwritten numbers. It fetches the previously trained neuronal network model from the Data Pipeline repository

Tensorflow MNIST ExampleInformation published on non-SAP site.
The graph can be separated into three main components:
  • Test image generation pipeline: Reads the images after generating and sends them to the evaluation operator.
  • Recognition pipeline: Loads a trained model, evaluates the images via tensorflow API calls and generates a json result output.
  • Stream UI pipeline: The Stream Recognition UI displays the json result and the image.

Prerequisites

  • (Only if run in local mode) Operators in this graph use the following libraries:
    • Python 2.7+

    • TensorFlow 1.7.0

    • SciPy 1.0+ & Pillow 5.0+

Configure and Run the Graph

Follow the steps below to run the training example from the Data Pipeline UI:
  1. In the left panel, select the Graphs tab and navigate to com.sap.ml.tensorflow.evaluateMnistUI.
  2. In the tool bar, select Run (play button).
  3. The Status panel indicates if the graph is running.
  4. Open the Terminal UI to see the accuracy and model save progress.