"yolov8-tfjs-vue-webrtc-demo"是一个利用深度学习目标检测模型YOLOv8、TensorFlow.js框架、Vue.js框架和WebRTC技术实现的演示项目。该项目可以通过前端界面实时展示视频流,并对视频中的目标进行实时检测和识别。借助GPU加速的WebGL后端,该项目能够实现快速高效的模型推理,同时通过WebRTC技术,还能够实现点对点的视频传输,为实时应用提供了强大的支持。"
"yolov8-tfjs-vue-webrtc-demo" is a demonstration project that utilizes the YOLOv8 deep learning object detection model, the TensorFlow.js framework, the Vue.js framework, and WebRTC technology. This project can display video streams in real-time through a front-end interface, and perform real-time detection and recognition of objects in the video. With the GPU-accelerated WebGL backend, this project can achieve fast and efficient model inference. Additionally, using WebRTC technology, it can enable peer-to-peer video transmission, providing powerful support for real-time applications.
link: https://yolov8-tfjs-vue-webrtc-demo.netlify.app/
将 YOLOv8 转换为 tfjs 模型,并将其放入 public 文件夹中。 打开 global.ts 文件并将以下代码块:
// if chrome version 113 above you can use webgpu
import '@tensorflow/tfjs-backend-webgpu'
export const modelName = 'yolov8n'
// replace
export const modelName = 'your model prefix name '
// import '@tensorflow/tfjs-backend-webgl';
编辑 labels.json 文件以匹配您的模型输出标签。
在终端中运行以下命令:
pnpm i && pnpm dev
- reference https://github.com/Hyuto/yolov8-tfjs