I build free, local-first AI for vision, LLMs and RAG, so your data never leaves your machine. Fast detection models, a deployment server you install with one command, and the tools in between. #Marseille with ❤️ love.
Every object detection library publishes a latency number, and almost none of them is the time your frame takes. We measured YOLO26, RF-DETR, D-FINE and mayaku end to end on one RTX 3060, from a 1080p frame to final boxes: some ~2 ms networks take 13 ms per frame, most of it spent on CPU pre-processing.
How we built CiliAI with two standard Mask R-CNN models and a handful of small-object tricks, and what I would do differently next time.
Fake vaccines leave different drying patterns than real ones. How we turned that into a two-model AI pipeline on 380 phone photos, and why I don't fully trust our 100%.
We will build a script that takes images from a camera stream, with a delay between them, at a given width in pixels and for a given duration. The images can come from a webcam or a video.
Finding faces in photos or videos is very cool, but it isn't enough information to build powerful applications. We need more information about the person's face, like its position, whether the mouth is open or closed, whether the eyes are open or closed, and so on.
In this post I will provide a bash script to install OpenCV 4.1.0 (with Python 3.6) on macOS Mojave and High Sierra. OpenCV 4 was released on 20 November 2018, with many bug fixes and other changes.