Skip to content
This repository has been archived by the owner on Nov 9, 2019. It is now read-only.

Home

Jump to bottom
raheelzubairy edited this page Nov 7, 2017 · 13 revisions

Short Name

Industrial Visual Analysis

Short Description

Analyze industrial equipment on sight for defects using Watson Visual Recognition in juncture with IBM Cloud Functions and Cloudant.

Offering Type

Emerging Technologies

Description

Performing industrial equipment inspection can be time consuming from hours to weeks, especially when equipment is all over the place for industries like oil and gas, transportation (roads and railways), construction and agriculture. This code pattern demonstrates how this inspection can be automated using images of the equipment to just point the user to the equipment needing attention to equipment meeting normal standards. The Watson Visual Recognition service is trained to identify particular defects or whether the image meets normal conditions. The analysis of the image is triggered by IBM Cloud Functions as an image is added to a Cloudant database.

Authors

Code

Video demo

Overview

In this code pattern, we will identify industrial equipment for various damages upon visual inspection by using machine learning classification techniques. Using Watson Visual Recognition, we will analyze the image against a trained classifier to inspect oil and gas pipelines with six identifiers - Normal, Burst, Corrosion, Damaged Coating, Joint Failure and Leak. For each image we will provide a percent match with each of the identifiers, on how closely the image matches one of the damaged identifiers or the Normal identifier. This data can then be used to create a dashboard to the pipelines needing immediate attention to no attention.

The images data is stored in a Cloundant database This code pattern demonstrates IBM Cloud Functions (OpenWhisk) to trigger microservice as an image is added to the Cloudant database. The microservice performs the Visual Recognition analysis and updates the Cloudant database with the analysis data.

When the reader has completed this code pattern, they will understand how to:

  • Train Visual Recognition to classify images
  • Configure Cloudant database to store and retrieve image data
  • Set up IBM Cloud Functions to trigger Visual Recognition analysis and store result in Cloudant database
  • Launch a web app to view a dashboard of the Visual Recognition analysis, and deploy to Bluemix

Flow

  1. User uploads the image through the web UI
  2. The image data is send to the Cloudant database
  3. As the image is inserted into the database, the Cloud Functions triggers mircoservice
  4. The microservice analyzes the image using the trained Watson Visual Recognition service
  5. The analyzed data is fed back into the Cloudant database
  6. The dashboard on the web UI displays the Visual Recognition analysis and images requiring attention

Included components

  • Visual Recognition
  • Cloudant
  • Cloud Functions

Featured technologies

  • Visual Analysis
  • Serverless
  • Database

Blog(s)

Industrial use case of Watson Visual Recognition

-by Raheel Zubairy

Industries have a lot of infrastructure. They range from thousands of miles of pipelines in the oil & gas industry to enormous acres of land in agriculture, and not to mention the colossal transportation structures such as roads, bridges and railway tracks. These structures, along with multitude of industrial equipment, need to be inspected and maintained for continuation of operation of these industries. Corporations employ various methods to do this such as routine visual inspections, running tests and digging deeper. However, with the number of structures and equipment to keep track, many inefficiencies exist.

A visual inspection can prevent disaster and maintain seamless operations. In most cases a quick look can identify whether there is a defect, and with some experience we can identify what the defect is. With Watson Visual Recognition service we can determine whether the equipment in the image meets normal conditions or by training the service to identify particular defects and damage. In our code pattern, Industrical Visual Analysis, we train the Watson Visual Recognition service to inspect oil & gas pipelines to classify the image into categories: Normal, Burst, Corrosion, Damaged Coating, Joint Failure and Leak. The training is performed by providing the service with a set of images for each category, by defining a particular classifier. This classifier can then analyze the image and provide a percent match to each category.

In our code pattern, we have setup the Visual Recognition analysis with IBM Cloud Functions and Cloudant database. When a user uploads an image, the initial image data is stored in our Cloudant database, which triggers a microservice using IBM Cloud Functions to analyze the image. Once the analysis is complete, the database is updated with the Visual Recognition analysis data. The pattern includes a web application which uses the analysis data to display a dashboard for the images needing immediate attention to images needing no attention.

Currently the web interface can be used by the user to upload images. However with the setup of Cloudant and IBM Cloud Function, this pattern can be extended to capture images from external sources and devices including drones, cameras on location, satellite images and more. So depending on the industrial use case, developers can use the Watson Visual Recognition service to automate the inspection process!

Related Links

Clone this wiki locally