Saving bugs, saving the world

Key Challenges

In their commitment to saving invertebrates and sustaining the world's ecosystem, the charity were looking to cut out the manual process of bug detection on vehicle registration number plates to analyse accurately the number of bugs caught on vehicles.

Key Details

Charity dedicated to the conservation of invertebrates

Our partner is a consultancy working with a charity based in the UK, dedicated to the conservation of invertebrates including bugs, insects, and bees. They are working towards a sustainable population of invertebrates, in order to sustain the world's ecosystem. One of the projects the charity is working on is analysing the number of bugs that were splattered on vehicle registration number plates to see the impact vehicles were having on the survival rate of a variety of bugs.

The manual process of bug detection

The existing model of bug detection was a manual process. Firstly, before the car journey, the number plate would need to be cleaned, once the car arrived at the destination, the bugs splattered on the number plate would be examined using a ‘splatometer’ grid. This process was labour intensive, therefore, they required a technology that would enable them to detect the object, in this case, bugs on the number plate after each journey.

Machine learning to automate the bug detection process

Based on a Machine Learning and Deep Learning model, Ardent's highly skilled software engineers developed sophisticated software that would detect the number of bugs on the vehicle registration plate. Using Cognito AI, the machine learning model was trained using over a hundred samples with YOLO (You Only Look Once) algorithm because of the unmatched benefits it offers in speed and accuracy.

"The biggest challenge that the team faced was that, due to the size of the bugs, it was difficult to train the ML model to differentiate between the bugs and dust."

Nikhil Singh

The software consisted of an easy to use User Interface (UI) which allowed the user to upload images for it to be processed with the machine learning/deep learning model utilising AWS technologies to deliver the results quickly and easily.  The results were also segregated by accuracy to enable the user to rule out low accuracy results.

More on our technology partners.

Proof of concept delivered successfully

Ardent’s software engineers delivered a successful proof of concept to the charity showcasing how they could track and monitor the number of bugs caught on vehicle registration plates easily, without manual processes to support their efforts on saving invertebrates.

"In order to overcome this, were able to crop the images, increasing the visibility of the bugs and consequently increasing the volume of data to further train the algorithm."

Nikhil Singh

Ardent object detection solutions

Object detection enables the user to locate objects in images and videos with AI and machine learning technology. It is used across industries ranging from autonomous driving and Advanced Driver Assistance Systems (ADAD) to video surveillance to healthcare. Ardent’s proof of concept of object detection proves how this technology can be game-changing in a wide variety of industries. If you are looking to employ AI and Machine learning for an object detection project, then get in touch to find out how we can help.