Recently, I saw that YOLO officially used YOLOv11, a pure vision-based approach, for speed estimation. I found it quite amazing because speed estimation is usually done using methods like radar and infrared. So, I did some research and discovered that the underlying principle is actually using vision to track the same target. The speed is calculated based on the distance the target moves between two frames (obtained by multiplying the number of pixels moved by the actual size of each pixel in real life) and the time per frame (usually 1/30th of a second, depending on the camera).
Then I do a project to show my result, and YOLOv11 inference on Hailo8 NPU, here is the link:https://github.com/Seeed-Projects/YOLOv11-Hailo-Tracker
![Image]()
Then I do a project to show my result, and YOLOv11 inference on Hailo8 NPU, here is the link:https://github.com/Seeed-Projects/YOLOv11-Hailo-Tracker

Statistics: Posted by jiahaoli — Wed Dec 17, 2025 8:32 am — Replies 1 — Views 28