Object Detection in Real time

• Clone the repo from https://github.com/SRai22/tfjs-object-detection.git
  

• cd tfjs-object-detection

• npm install

Running with Wasm backend

    npm start 

• Opens the browser, allow the browser to access camera, click the button to start the detection.

Running with WebGPU backend

• Download and Install Chrome Canary from https://www.google.com/chrome/canary/

• Open chrome, and goto: chrome://flags/ , enable the option #enable-unsafe-webgpu

• Relaunch chrome

• Goto tfjs-object-detection/src/components/ObjectDetection.js --> edit line

66           await detector.current.setModelBackend("wasm");

                        to 

66           await detector.current.setModelBackend("webgpu");

    npm start

• Open with Chrome Canary, allow the browser to access camera, click the button to start the detection.

TODO

• Parameterize the backend variable