Luís Condados·Aug 20CNN vs Transformer, quantized: YOLO26-seg and RF-DETR-Seg race for instance masks on an Intel iGPUInstance segmentation, a mask per object rather than just a box, is the expensive end of real-time vision. The reflex is to reach for an…
Sobhan Hota·Jun 22Building a Real-Time Human Pose Detection App with RF-DETR and StreamlitPose detection sounds simple — find a person, mark their joints. But in practice it falls apart fast:
Davidshuaiche·Jun 9ONNX Runtime Go Binding: Zero-CGO Calls via puregoHow raven-onnxruntime uses purego to call the ONNX Runtime C API from Go, avoiding CGO dependencies and simplifying cross-compilation and…
Luís Condados·May 29RF-DETR vs YOLO-NAS: A Practical Benchmark for Edge Deployment on CPU, GPU, and Intel iGPUWhen accuracy matters, pick the transformer. When latency matters, pick the CNN. This article gives you the numbers to defend that call.
Daisuke Majima (MLBoy)·Mar 31Filling the “Last Mile” of Object Detection — peaceofcake 🍰 as an OptionWhat happened in object detection in 2024
