IoT Analytics Lifecycle Use Cases
for developers


Anomaly Detection in Floodlights for Smart Campus
Learn how to build a model for real-time detection of malfunctioning light groupings using SAS IoT analytics.

Key takeaways from the example: 
 • 
design streaming model for real-time failure detection
 • use subspace tracking algorithm to detect anomalies
 • best practices for subspace tracking algorithm

>> Resources on GitHub <<

Show details

Overview

Learn how to build a model for real-time detection of malfunctioning light groupings. These light groupings are part of a Smart Campus floodlight circuits located in a parking lot.

The data set consists of energy consumption values captured every five minutes from six floodlight circuits, over a span of about three months. 

We will use the subspace tracking (SST) algorithm packaged in SAS Event Stream Processing Studio to detect outliers in real time using streaming data. It is a method to detect anomalies and degradation in systems that generate high-frequency, high-dimensional data. It can be used for data containing a single measure for sensors from various devices operating under similar conditions (e.g., energy output from multiple panels in a Solar farm), or multiple measures for sensors from a single device operating under similar conditions (e.g., turbofan in an aircraft).

Tags

CategoryCondition-based maintenance
Sub-CategoryAnomaly detection
Analytical MethodSubspace tracking (SST)

 


Demographic Image Analysis
Learn how to train and build a computer vision model for demographic detection using SAS IoT analytics.

Key takeaways from the example: 
 •
cover basics of Computer Vision (CV)
 • train CV models using the SAS Deep Learning Python (DLPY) package
 • operationalize our models using streaming analytics

>> Resources on GitHub <<

Show details

Overview

Imagine you are the owner of a retail store and want to better understand your customers. Wouldn't it be great to know what percentage of customers in your store looked happy as they browsed your shelves, or perhaps which aisles attract men versus women customers?

In this example, we used two products that are embedded in the SAS Analytics for IoT solution, SAS Visual Data Mining and Machine Learning (VDMML) and SAS Event Stream Processing (ESP) to show you how to train and deploy various demographic analytic models to create a single solution that connects directly to a USB-connected security camera. With this connection, we can process incoming images using pre-training analytical models and produce demographic statistics.

This application provides:
     •  Face Detection - process incoming images and detect human faces in the video stream
     •  Age Classification - determine approximate age grouped by category
     •  Gender Classification - determine male or female
     •  Emotion Classification - group by these classes: Happy, Neutral, Sad, Fear, Surprise, Angry

Tags

CategoryComputer Vision
Sub-CategoryImage Classification
 Object Detection
Analytical MethodResNet-50
 Tiny YoloV2

Detecting Degradation in Wind Turbines
Learn how to monitor degradation of wind turbines in real-time using  SAS IoT analytics.

Key takeaways from the example: 
 • 
design streaming model for real-time failure detection
 • use subspace tracking algorithm to detect anomalies
 • best practices for subspace tracking algorithm

>> Resources on GitHub <<

Show details

Overview

Learn how to monitor degradation of wind turbines in real-time. In this example, we will use simulated data which is the hourly energy (in kilowatts) that each of the four turbines produces.

We will use the subspace tracking (SST) algorithm packaged in SAS Event Stream Processing Studio to detect degradation in real time using streaming data. It is a method to detect anomalies and system degradation in systems that generate high-frequency, high-dimensional data. It can be used for data containing a single measure for sensors from various devices operating under similar conditions (e.g., energy output from multiple panels in a Solar farm), or multiple measures for sensors from a single device operating under similar conditions (e.g., turbofan in an aircraft).

Tags

CategoryCondition-based Maintenance
Sub-CategoryAnomaly Detection
Analytical MethodSubspace tracking (SST)

Image Classification Using Rubik's Cubes
An image classification example that uses a Rubik's Cube to demonstrate how to apply the the technique in manufacturing quality using SAS IoT analytics.

Key takeaways from the example: 
 •
use SAS Analytics for IoT to train and build an image classification model
 • train CV models using the SAS Deep Learning Python (DLPY) package
 • operationalize  models using streaming analytics and ESPPy

>> Resources on GitHub <<

Show details

Overview

Imagine you are responsible for manufacturing quality at your company. Wouldn't it be great to know what percentage of product contains flaws? Maybe replace human oversite with automation or collect data to perform a root cause analysis. SAS Analytics for IoT is a perfect solution to solve this or any Computer Vision (CV) problem. This comprehensive AI-embedded solution provides so many capabilities because it integrates a number of key SAS products.

In this use case, we are going to simulate our manufacturing process using a Rubik's cube. Our cube represents a part which is being produced, and it is up to us to determine whether it was manufactured correctly.

Tags

CategoryComputer Vision
Sub-CategoryImage Classification
Analytical MethodResNet-50

Anomaly Detection in Air Handling Units
Learn how to build an anomaly detection model and deploy it for real-time monitoring of malfunctioning Air Handling Units (AHUs) using  SAS IoT analytics.

Key takeaways from the example: 
 •
build an anomaly detection model using Support Vector Data Description (SVDD) algorithm
 • deploy an offline model for real-time failure detection

>> Resources on GitHub <<

Show details

Overview

Learn how to build an anomaly detection model and deploy it for real-time detection of malfunctioning Air Handling Units (AHUs). These AHUs are part of a smart campus heating, ventilating and air-conditioning (HVAC) system.

We will use the Support Vector Data Description (SVDD) algorithm, packaged in  SAS Visual Data Mining and Machine Learning (VDMML), and deploy it in SAS Event Stream Processing Studio (ESP) to detect outliers in real time using streaming data.

Tags

CategoryCondition-based Maintenance
Sub-CategoryAnomaly Detection
Analytical MethodSupport Vector Data Description (SVDD)

Tracking Social Distancing Using Computer Vision
Identify objects (i.e., people) who are following social distancing guidelines and those who are not using  SAS IoT analytics.

Key takeaways from the example: 
 •
use Tiny Yolo V2 model to detect people in images
 • calculate the distances between all objects
 • determine whether we have crowds in our image

>> Resources on GitHub <<

Show details

Overview

Due to the recent pandemic, several governments have implemented social distancing measures to limit the spread of the disease.  In this use case, we show you how to track these social distancing rules using Computer Vision on camera image.

The overall goal of this project is to identify objects (i.e., people) who are following social distancing guidelines and those who are not. This is accomplished by way of the following:
  •  Transform images and object coordinates into a two-dimensional map
  •  Calculate real world distances between detected objects 
  •  Detect crowds that exceed a specified parameter
  •  Visualize the results on the camera image

Tags

CategoryComputer Vision
Sub-CategoryImage Classification
 Object Detection
Analytical MethodTiny YoloV2
 cKDTree
 query_ball_tree

Anomaly Detection from Mobile Data
Learn how to collect live accelerometer data from a mobile device and pass it to ESP using an MQTT protocols with SAS IoT analytics

Key takeaways from the example: 
 • 
JSON data is decoded and passed to real time trained models k-means and SST
 • compare new observations against baseline data and any deviation will be considered an anomaly
 • provides ability for machines  to self-train their own unique anomaly detection models

>> Resources on GitHub <<

Show details

Overview

This project will show how to collect live accelerometer and gyroscope data from a mobile device and pass it to Event Stream Processing (ESP) using an MQTT protocol. Once in ESP the JSON data is decoded and passed to the unsupervised machine learning models k-means and subspace tracking (SST). These unsupervised machine learning models are able to find patterns in data and are trained dynamically from streaming data.

In this use case we consider the mobile device as a sensor placed on any stationary piece of equipment on a factory floor. For demo purposes our anomaly detection time frame is 30 seconds. In a real world scenario, this time frame would be much longer. That is to say as long as the equipment continues to generate 30 seconds of consistent readings, this will be considered the baseline. Once that 30 second baseline is established, new observations are compared against it and any deviation will be considered an anomaly. KPIs determine the severity of the anomaly. With this type of self-training the baseline is dynamically set and does not have to be calculated separately for each piece of equipment. If the sensor is placed on a rotating object, the accelerometer data becomes the norm, as long as it is consistent.

Imagine a factory floor with 1000 machines, each oscillating at different amounts. Now imagine how much time would be saved if the sensors placed on these machines were able to self-train their own unique anomaly detection models. Very powerful.


Bird Watching
Learn how to train a neural network model with SAS ESP by capturing bird images.

Key takeaways from the example: 
 •
train a neural network model using SAS VDMML to identify birds
 • have SAS ESP process and score images
 • have SAS ESP email us with the bird's picture and predicted species

>> Resources on GitHub <<

Show details

Overview

Bird watching is a great and relaxing hobby, and by getting a bird feeder, we can enjoy as the birds come by during the day for a little snack. It's also nice to take pictures of the birds so we can find out the species and show them to our friends. Taking pictures of the birds requires us to have our phone or a camera handy whenever watching the action at the bird feeder, and being able to take a picture without scaring the birds away. Also, what if some great birds stop by for a nibble and we aren't around?

To address these issues, and bring the IoT to bird watching, we created a fun project with SAS software to make sure we never miss the action at our bird feeder.

Tags

CategoryComputer Vision
Sub-CategoryImage Classification
Analytical MethodConvolutional Neural Networks (CNN)

Back to Top