The Eye of AI #9

Computer vision is a truly transformational technology for the retail industry, but it has applications that touch nearly every aspect of our lives. This is our weekly summary of what happens when cameras, computers and great ideas get together.

Hardware for Hardware Recycling

Gen X kids will surely remember how robots used to freak out about disassembling. Well, Microsoft's new "recyclobot" uses machine learning and computer vision to disassemble hard drives and recycle 90% of their components, focusing on valuable materials like neodymium magnets. While data centers typically shred millions of drives annually, Microsoft's robot offers a more sustainable solution by automating the disassembly process. This project, part of a Global Hackathon, aims to achieve a 90% reuse and recycling rate by 2025. Though still in development, it represents a promising shift towards eco-friendly e-waste management.

Seeing Like Humans

People assume that the things they see are a pretty good representation of reality. Sadly, that's not the case. Our brains use a lot of energy to process many parallel inputs while also generating thoughts and ideas, so the stream of visual information is treated with a lot of generalisation and inference rather than laser focus on every in-coming photon. This causes an alignment problem between AI models and human experience. With this in mind, Google DeepMind researchers have proposed "AligNet," a framework that aligns AI vision models with human visual perception. By simulating human-like similarity judgments and incorporating uncertainty measures, AligNet improves the alignment of machine representations with human cognition. The framework enhances AI models' generalization and robustness, showing significant performance gains across various tasks involving object similarity.

Small Town Tech

While tech giants look for ways to engage mega-conurbations with their Smart City technologies, many smaller councils and local authorities are turning to piecemeal AI systems to give them insights into the populations they serve. As an example of this, Ashburton, a town of 21,000 inhabitants on New Zealand's South Island has recently installed two computer-vision pedestrian counters in its town centre to monitor foot traffic. Positioned near two pedestrian crossings in the town centre, the counters provide anonymous data on pedestrian patterns, helping both the council and local businesses with planning. These counters, which also track rainfall and temperature, will provide updates at the Council’s Activity Briefings every six weeks.

How Sweet it Is

As any avid gardener knows, catching a strawberry in a state of perfect ripeness is what summer is all about. So Researchers at Western University have developed an AI and computer vision system to detect strawberry ripeness and disease with near-perfect accuracy (up to 99%). This open-source solution, designed for indoor and outdoor farming, helps farmers monitor crops in real-time and reduce food waste. By using synthetic images and requiring minimal data, the AI system is accessible to farms of all sizes, offering a cost-effective and scalable tool for precision agriculture. The software is free for farmers to adapt to their specific needs. Now if they can figure out a way to stop the birds getting there first...

Enhancing Fall Detection

As populations age there is a growing market for solutions that help people live independently for longer. Researchers in Saudi Arabia have developed a new fall detection system with improved mechanisms for accuracy and speed. This system uses a combination of deep learning models and advanced optimization techniques to monitor and detect falls in assisted living environments. The approach improves accuracy by filtering out noise and enhancing real-time detection, which helps ensure the safety of elderly residents. The method demonstrated superior performance compared to existing models, though further testing in varied conditions is needed.

That's everything for this week. Please keep an eye on the SAI Group blog for everything that we're thinking and talking about, including a new article this week about how the Co-operative Group is tackling crime in their stores.

Got some cool tech to share? Whether its your own project, or just something you saw and thought "I want people to know about this!", let us know about it and we'll include it in upcoming editions.