On the 7th of April, Newcastle University Business School and Pitch-In Project co-hosted the second IoT Newcastle event, which is typically held every first Tuesday of the month. Sponsored by www.goto50.ai and Newcastle University, IoT meet-up is a platform for discussions around IoT and AI developments and implementation. It gathers people who are involved in the IoT ecosystem and interested in the adoption of technology or looking for solutions to the delivery of automation and intelligence.
The April meet-up had a talk from Oliver Hamilton, Director of Computer Vision at COSMONIO (https://www.cosmonio.com/). COSMONIO is a software development company developing AI-based systems focusing on computer vision and image processing. COSMONIO designs AI systems extracting visual information from images based on self-learning algorithms. The company runs research and development labs focusing on intelligent vision systems, medical technologies based on deep learning and computer vision for life sciences.
Oliver Hamilton presented an NUOS machine learning interactive platform that works without writing a single code. The platform automates visual inspection tasks, such as the detection, segmentation and classification of instances on images or video, identification of anomalies, counting and localisation of instances. Oliver demonstrated how the system works when it is applied to object detection. For the system to build a dataset, a user needs to capture and annotate objects. This can be done by drawing bounding boxes on images or video or by using an AR kit to capture a 3D model of the object in space. The generated dataset is used by the system for self-training, which might take several rounds. Each round of training requires edits and feedback from the user to improve the task implementation through incremental self-learning. Oliver also explained how the platform can be deployed to edge computing devices.
The all-inclusive NUOS system is capable of solving complex pattern-recognition problems in the fields of biology, health, defence, industrial inspection and space exploration. For example, NUOS can be a time-, labour- and cost-efficient solution for detecting road traffic offences captured on outdoor cameras. By tasking the deep-learning machine, the system can be trained to detect and capture only the anomalies, without the need to record the entire chain of events. The deployment of the system can save time spent on manual data inspection, and it reduces the amount of recorded data and associated costs.
To participate in future IoT meet-ups and learn more about how advances in AI and IoT can benefit organisations in different sectors, follow the schedule of events here: https://ebusiness.ncl.ac.uk/projects/pitchin-iot-ecosystem/iot-meetup.php.