Actionable Insights at Scale and Speed

We combine human expertise with AI to improve process efficiency, reduce costs and improve outcomes. 

Scale Processes and Reduce Costs with ML and AI

We collect and process imagery at scale and speed; extracting objects of interest as well as determining their state. This enables customers to reduce costs, massively scale what they do and unearth hidden insights and knowledge.

Our technology is being used for condition assessment of utility assets, for generating ecological surveys of transport corridors and assuring the outcomes of innovative manufacturing processes.

Our unique approach to onboarding clients ensures delivery of business outcomes within 7 weeks.


Latest Applications

Manufacturing Quality Assurance

Leaks in a hollow silicon membrane are identified by passing air through the fibre and submerging it. Bubbles appear at the site of an imperfection.

Keen AI developed fast computer vision based tools to highlight areas in the scene where leaks are present and present to a human who can repair the leak. The factory environment is very noisy with the additional complication of uneven lighting.

Condition Assessment

The Keen AI platform (KAI) extracts key components from footage collected by helicopters or drones and presents them for review to a condition assessment engineer.

This video shows the application of the background deep learning model used to extract components being applied to a sample video of a single conductor suspension tower in real time.

Corrosion Detection

Corrosion is an ever present issue for electricity transmission and distribution assets. Keen AI combined traditional computer vision techniques and convolutional neural networks in order to identify candidate regions where corrosion maybe present.

Bespoke edge detection algorithms are used in a follow up step identifies regions of the image likely to contain steel and thereby separate background from tower.

Ecological Surveys

We have developed a vehicle mounted solution for collecting high resolution imagery of roadside vegetation. AI is then used for detecting the presence of invasive plant species or Ash trees.

This provides a rapid, high quality vegetation survey methodology, resulting in cost and time savings for our customers. 

Latest News & Updates

Reduce Manual Assessments and Achieve Automation with KAI Machine Learning.

Our AI platform, KAI, sees the world, and learns. Whether your goal is to augment or automate a process, KAI makes this easier. By learning from your inputs, KAI helps you continually reduce manual effort in identifying objects of interest in images and video, driving massive resource efficiencies for your organisation or business.

Increase process productivity

Support human experts by rapidly identifying objects of interest, increasing process productivity by either telling you where to look, or skipping what you don’t need to look at. Clients are recording a 60% reduction in process completion time.

KAI learns from human experts

KAI continually learns from assessments made by human experts. Over time, reinforcement learning techniques are used to predict the assessment that a human will make. Once predictions reliably mimic the human expert, the process can be automated.

Easy to deploy deep learning models

A simple interface for creating and deploying deep learning models. Intuitively designed to remove the complexity of machine vision, ML and AI.

Assessment standardisation

Standardise process outcomes and remove inconsistencies by augmenting your process with AI. Data collected by KAI can be used to standardise process outcomes and create an ideal human expert for KAI to mimic.

Easy to integrate, in the cloud or on premise

KAI is easy to integrate with your applications for training and deploying models for new scenarios. KAI can be delivered as a cloud based application, delivering complete scalability for high demand, or installed as an instance on your internal networks.

Unlock insights from years of footage

KAI can analyse everything helping expedite automation, if that is the goal. Historical footage, live feeds, thousands of hours, unlimited imagery all improve learning and accuracy for your application. Unlock years of footage for asset inventory, and make content searchable.

Increase Efficiency, Improve Outcomes and Reduce Costs with ML

National Grid UK - Reduction in time taken to process video footage
OxyMem - Reduction in time taken to process video footage
Pilot - Automatic detection of corrosion on steel work
Conducting AI Ecological Surveys - Reduction in cost compared to traditional methods

National Grid - Spacer Detection

National Grid were able to achieve significant time savings by identifying and extracting components of interest for review.

Multiple Component Detection

National Grid were able to achieve a 66% reduction in the time taken to review over head line footage for condition assessment, and are now training a machine learning algorithm to automatically identify defects.

National Grid - Detecting Corrosion on OHL Tower Components

National Grid were able to automatically remove 25% of towers from the inspection process by identifying clean towers with no corrosion.

Machine Assisted Quality Control

Oxymem needed a cost-effective solution to standardise a visual QA process, that could be deployed quickly, without the need for expensive equipment, or a tightly controlled environment.

OxyMem - Machine Assisted Quality Control (Stage 2)

KAI detecting bubbles from a leak with heat-map overlay showing leak persistence (slow-motion).

Combining AI with your process expertise can be immensely powerful in delivering efficiencies, improving outcomes and reducing costs. 

Talk to us about your business process today and discover how AI can help.

DScience Ltd © 2021 | The Moseley Exchange, 149-153 Alcester Road, Birmingham. B13 8JP