The Eye of AI #6

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.

Stop Trains Deer-railing

Deer account for more human deaths than wolves, bears, sharks and snakes combined. There are 2.1 milllion vehicle-deer collisions a year in the US, causing 60,000 injuries and over 400 deaths. In the UK, 700 people are killed and injured by deer. The problem isn't limited to motor vehicles - stupid deer also stroll along railway lines and stop and stare at the on-rushing headlights. Network Rail, the UK's rail track operator, has deployed cameras along the lines to detect and scare away rogue deer, tracking their movements to ensure they stay away from trains. They estimate they have saved the lives of 6,000 deer so far, ironically increasing the chances of collisions with road traffic, given the recent infrequency of rail services.

Archaeology AI

We all know that Mount Vesuvius exploded about 2,000 years ago, freezing the citizens of Pompeii in a particular moment for the rest of time. Of course, the damage and devastation covered a much wider area than a single ancient city, with ash and rock being spread far and wide. At was has become known as the Villa dei Papyri, an entire library (initially mistaken for fishing nets) has been uncovered - one of the largest in the ancient world. And of course it was only a matter of time before archaeologists turned to computer vision and AI to decipher the ravaged rolls. So far, the technology has uncovered works by Philodemus including the books 'On Flattery', 'On Arrogance', 'On Household Management', and 'On Death'.

Meta gets more, well, meta

Ever wondered what it would be like to slice yourself up into a million-million pieces and reassemble inside a computer game or virtual simulation? Well, wonder no more because Meta's Sapiens model has arrived, with broad applications due to its advanced human-centric vision capabilities. In AR/VR, it enhances realism by providing accurate human pose estimation and detailed body-part segmentation, leading to more lifelike avatars and interactions. In healthcare, sports analytics, and security, Sapiens can analyze human movements with precision, aiding in diagnostics, performance improvement, and monitoring behavior in crowded spaces. The model also plays a significant role in improving gesture and facial recognition systems in human-computer interaction.

Tomay-to, Tomah-to

It's the time of year where many of us hustle around the garden, harvesting this year's crop of sweet, delicious salad fruits to eat now or preserve in jars to brighten the winter days. Wouldn't it be nice to have a robot do the backbreaking work, though? Well, researchers at Shandong Agricultural University in Tai’an, China are working out ways to make this possible. To solve the problem of tomatoes playing hide-and-seek behind leaves, stems and shadows, S-YOLO, a lightweight yet powerful model, turbocharged with smart algorithms and attention modules, has been able to spot even the most elusive tomatoes with 96.6% accuracy and lightning-fast speed. By slimming down the traditional YOLO model and beefing up its detection prowess, S-YOLO not only outshines its predecessors but also promises to make tomato-picking robots more efficient, transforming the way we harvest our backyard bounty.

Harvesting Solar Efficiency

MIT engineers have developed a computer vision technique that dramatically speeds up the process of evaluating new materials for solar cells and other devices. By automatically analyzing semiconductor samples, this AI-powered method estimates critical properties like band gap and stability in minutes—tasks that would take humans days to complete. The goal? An autonomous lab where AI predicts, creates, and tests materials around the clock, pushing the boundaries of solar cell efficiency and beyond. It's a bright future for renewable energy, fueled by the power of automation and AI.

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 to activate your passive loss prevention strategy.

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.