Roboflow

🪲 Video: Unlocking big data research at museums with vision AI How do you digitize 8 million insect specimens and turn them into usable scientific data? Last week I spoke with Elizabeth Postema, Postdoctoral Researcher at the Field Museum of Natural History, about how she and other researchers created a vision AI solution to answer that question. See the full interview here: https://lnkd.in/gN-KNwqR We explored several topics: 🐞 How the solution uses vision AI to automate specimen imaging and analysis, like measure body size and transcribing physical labels 🐞 Examples of big data research opportunities that were unlocked by digitizing insect specimens (e.g., verifying Bergmann's Rule) 🐞 Best practices for developing custom AI models and the tools in Roboflow that sped up the process

I recently shipped a computer‑vision demo in 60 minutes - and there’s a $10M reason hospitals should care. Most teams still juggle fragmented tooling (label here, train there, deploy elsewhere). I wanted to show what end‑to‑end really looks like - so I built a kit inspection model using Roboflow. What it does: - Identifies each tool - Flags small defects - Detects missing items - Marks the kit pass / fail Time to live demo: ~1 hour (Yes, that includes labeling, dataset optimization, training an RF‑DETR model, Workflow logic, and deployment - all with Roboflow). Why? A two‑hospital campus analysis estimated $6.75M-$9.42M/year from instrument‑related delays. Examples like these can be remedied with computer vision. We can see impact in things like lab inventory management, fewer reprocessing loops, as well as audit trails supporting AAMI ST79 compliance.

🚘 Audi Reader: Helping people understand their vehicles with vision AI I recently had the pleasure of interviewing Lorenzo Spadoni from OgilvyES about the development process behind the Audi Reader application. The application is powered by custom AI models, allowing users to point their phone at a feature in their vehicle -- like an unknown button or indicator light -- and instantly get more information. In the video, we chat about: 🚙 The motivation behind developing Audi Reader (hint: 70% of drivers don't read the car user manual) 🚙 How model training data was captured with DSLRs vs. smartphones 🚙 Improving performance by combining object detection and classification models in Roboflow Workflows Watch the video here: https://lnkd.in/gyKMBmbw