YOLOv11 With Depth Camera on Jetson Orin For Distance Measurement | Seeed Studio Wiki

[giscus[bot]]

YOLOv11 With Depth Camera on Jetson Orin For Distance Measurement | Seeed Studio Wiki

This guide shows how to pair an Orbbec Gemini 2 depth camera with a Jetson‑based reComputer (JetPack 5.1.3) to perform real‑time object detection and 3‑D distance measurement. It walks through installing the Gemini 2 Python SDK, building it with pybind11, and validating camera output via an OpenCV test script. Next, it details converting a YOLOv11n model to a TensorRT engine for high‑speed inference on Jetson, then explains how to retrieve camera intrinsics and project depth pixels into 3‑D coordinates so that Euclidean distances between detected objects can be computed. A sample script combines YOLOv11 detections with depth data to display measured distances (typically accurate to ±1 cm), and the article closes with tips—depth‑map filtering, centering targets, and frame averaging—to mitigate structured‑light depth noise on edges, slanted or reflective surfaces, and under challenging lighting.

https://wiki.seeedstudio.com/yolov11_with_depth_camera/

[ravitol]

Can this setup be extended for ball tracking scenario and also line calling using real time edge.

[zibochen6]

@ravitol Precisely locating and rapidly moving targets can be addressed by applying target tracking algorithms such as Kalman filtering under the original target detection algorithm, to ensure real-time tracking without losing the target.