Can Images be a new source of useful information

3 min read

Predictive modelling solutions have relied on the availability of good quality, historical data and models that are estimated on representation of past behavior. For example, credit scoring to predict propensity of customer default requires the use of clean, processed transactional and demographic data. Datasets used for this type of analysis are first captured through conventional channels, such as forms and third party databases, and then processed in batch. This process assumes that scoring and alert generation is not time-critical. In other words, it is assumed that there is sufficient time (hours rather minutes or even seconds) before a relevant action for an alert expires. It also assumes that the data required regarding individual or entity being assessed is available through conventional data gathering exercises, such as a bank form submitted by the individual, agents or third party.

Whilst in credit scoring these assumptions hold true, there are situations where such assumptions are unrealistic. Risk assessments by customs agencies, border control, front-line policing operations, and even anti-money laundering operations are all time sensitive situations wherein the necessary action from the analytics assessment expires much more quickly (in minutes or seconds). . In addition, the data required to assess risk are not always collected through conventional methods. For example, a terrorist will not submit a form with their details. Similarly, travellers with goods entering a controlled customs area aren’t likely to submit a long form providing details of their contraband. Individuals’ expressions, movements, the tone of their voice and even body language all carry valuable clues which can considerably enhance subsequent risk assessments.

Image processing and related artificial intelligence (AI) models can considerably enhance the available data to assess risk in real time. The rest of this blog explains how:

SAS® Viya™ is a cloud enabled, in-memory platform that can train low-level images using various pre-processing and hierarchical representation methods (for a syntax example see figure 1). Andrew Pease gave a good overview on how to achieve this with great use cases in his recent blog (a visual example is presented in figure 2 and 3).  Once images have been trained, classification can be made from test datasets, with the resultant, expected outcomes fed into a real time decision flow.

Figure 1: Model specifications using SAS Viya actionset in Jupyter Notebook

Figure 2: Pedestrian detection

Figure 3: Feature extractions from images

An alert is generated to flag if someone should be considered a threat or not. Apart from obvious merits, this approach minimizes false positives which could be costly both to the parties involved and the entity.

Once the real time decision flow has been defined, web services can be triggered immediately generating alerts to notify stakeholders if an activity is estimated to be genuine or not.

What are the business benefits?

Often, analytics defined by data science teams are not optimally utilized by an organisation to achieve their highest business value. Building complex algorithms without first setting business goals defining how to implement such models leads to production delays that depreciate model relevance wasted opportunities. Dr. Steven O’Donoghue, in his recent blog, details how to set up a high-performing analytics team within an organisation for success.

For many use cases, the benefit of augmenting existing datasets with image data can be tremendous. Time sensitive situations, when either traditional transactional data aren’t readily available or perhaps not as descriptive of events as needed, can be vastly improved when images are added to the data mix. Image could be used to improve the existing data, decrease errors rates of results (e.g. false positives and potentially contain a plethora of insight not previously used. In banking, image processing for example can be used with existing datasets to screen loans and mortgage applications, recognize fraudsters using ATM machines. Custom agencies could be better equipped to prevent individuals with criminal records from smuggling hazardous items into the country. Images streaming from airport locations, boarders and dockyards CCTVs can be used to keep countries safe.

How would use image data within your organisation to improve upon your existing business?

About Larry Orimoloye

Larry has experience in the design and delivery of analytics solutions in the telecom, technology, oil and gas, pharma, banking and logistics sectors, among others. He has helped clients establish centres of excellence with an analytics remit across the organization, and designed and implemented customer-centric, real-time decision platforms using a combination of statistics, big data and machine learning techniques. Larry holds a master’s degree in applied statistics and data mining from the University of St. Andrews, Scotland.