Skip to content

Home

Jump to bottom
Olivier edited this page May 23, 2024 · 4 revisions

Welcome to the ImageAnalysis wiki!

Main Screen

The tool starts with a FasterRCNN object detection model loaded by default. To check everything is ok just load any image from your computer and click on the Detection On button to see a first result, as shown above. You can also click on the Camera button to perform analysis on your computer webcam feed, but note that this default model is not that fast, it lasted 1.5 secs on my Mac Mini M2 to perform its inference and display the results on an image, so the analyzed camera feed will actually display quite slowly.

To choose any other model from Torch hub just choose any other from the top menu bar under object detection, image segmentation or image classification menu. Should you prefer to test a model from Huggingface hub :

  1. just go to their model repository. Select any model you like, do not hesitate to use the filters from the left panel (at least for browsing the proper computer vision category)
  2. on the model page, copy the model name (in a form like author/model_name - note that a convenient copy button is available just near the model name)
  3. then, go back to our application, select the proper model type from the HuggingFace Models top menu
  4. and just paste the model name into the popup window asking for it.
Clone this wiki locally