Computer Vision inference models for Edge 2

Showcase
1

go-rknnlite is a project providing CGO bindings to Rockchips rknn-toolkit2 for the Go programming language. It allows you to run various Computer Vision inference models on the RK35xx series of SoC, such as on the Khadas Edge 2.

A number of examples are provided covering the following Models:

  • Image Classification (MobileNet, Pooled runtime)
  • Object Detection (YOLOv5, YOLOv8, YOLOv10, YOLOv11, YOLOX, RetinaFace)
    Object Detection
  • Instance Segmentation (YOLOv5-seg, YOLOv8-seg)
    Segmentation Sample
  • Pose Estimation (YOLOv8-pose)
    Pose
  • Oriented BBoxes (YOLOv8-obb)
    OBB
  • License-Plate OCR (LPRNet, ALPR combining YOLO+LPRNet)
    License Plate
  • Text ID (PPOCR detect, recognise, system)
    Detect
  • Streaming + Tracking (HTTP video with YOLO + ByteTrack)
    https://youtu.be/M6mvHTNQZqM
1 Like